Other Parts Discussed in Thread: AM69A
Tool/software:
I downloaded model from https://github.com/TexasInstruments/edgeai-tensorlab/blob/main/edgeai-modelzoo/modelartifacts/AM69A/8bits/ss-8750_onnxrt_ade20k_hf-transformers_segformer_b0_finetuned_ade_512_512_simp_onnx.tar.gz.link
then i ran infer on sk-am69a and i got error
root@am69-sk:/zxseg# python3 infer_test.py
libtidl_onnxrt_EP loaded 0x3ff12eb0
Final number of subgraphs created are : 1, - Offloaded Nodes - 396, Total Nodes - 396
Segmentation fault (core dumped)
so i compile on my own
method one :
/root/ti2/edgeai-tidl-tools/examples/osrt_python/ort/onnxrt_ep.py
"ss-ort-nvidia-b0": create_model_config(
task_type="segmentation",
source=dict(
model_url="",
infer_shape=True,
),
preprocess=dict(
resize=512,
crop=512,
data_layout="NCHW",
pad_color=0,
resize_with_pad=False,
reverse_channels=False,
),
session=dict(
session_name="onnxrt",
model_path="/root/ti2/edgeai-tidl-tools/kenny/test_model/segformer_b0_finetuned_ade_512_512_simp.onnx",
# meta_arch_type=3,
input_mean=[123.675, 116.28, 103.53],
input_scale=[0.017125, 0.017507, 0.017429],
input_optimization=True,
),
postprocess=dict(with_argmax=True),
extra_info=dict(num_images=numImages, num_classes=150),
),o/p
root@b7f0ab54a02d:~/ti2/edgeai-tidl-tools/examples/osrt_python/ort# python3 onnxrt_ep.py -c -m ss-ort-nvidia-b0 Available execution providers : ['TIDLExecutionProvider', 'TIDLCompilationProvider', 'CPUExecutionProvider'] Running 1 Models - ['ss-ort-nvidia-b0'] Running_Model : ss-ort-nvidia-b0 Running shape inference on model /root/ti2/edgeai-tidl-tools/kenny/test_model/optimized_model.onnx ========================= [Model Compilation Started] ========================= Model compilation will perform the following stages: 1. Parsing 2. Graph Optimization 3. Quantization & Calibration 4. Memory Planning ============================== [Version Summary] ============================== ------------------------------------------------------------------------------- | TIDL Tools Version | 11_00_06_00 | ------------------------------------------------------------------------------- | C7x Firmware Version | 11_00_00_00 | ------------------------------------------------------------------------------- | Runtime Version | 1.15.0 | ------------------------------------------------------------------------------- | Model Opset Version | 17 | ------------------------------------------------------------------------------- ============================== [Parsing Started] ============================== [TIDL Import] [PARSER] WARNING: Network not identified as Object Detection network : (1) Ignore if network is not Object Detection network (2) If network is Object Detection network, please specify "model_type":"OD" as part of OSRT compilation options [TIDL Import] WARNING: Resize layer - /decode_head/Resize_3 with scales > 4 is not optimal ------------------------- Subgraph Information Summary ------------------------- ------------------------------------------------------------------------------- | Core | No. of Nodes | Number of Subgraphs | ------------------------------------------------------------------------------- | C7x | 382 | 3 | | CPU | 12 | x | ------------------------------------------------------------------------------- ------------------------------------------------------------------------------------------------------------------------------------------------------- | Node | Node Name | Reason | ------------------------------------------------------------------------------------------------------------------------------------------------------- | Split | _token_0 | Layer 295 - op type Split, Unknown input dimension, not supported by TIDL | | Transpose | Transpose | Layer 296 - op type Transpose, Unknown input dimension, not supported by TIDL | | Squeeze | 0_squeeze_1 | Layer 299 - op type Squeeze, Unknown input dimension, not supported by TIDL | | Transpose | Transpose_token_3 | Layer 303 - op type Transpose, Unknown input dimension, not supported by TIDL | | Squeeze | 0_squeeze_0 | Subgraph does not have any compute node | | Transpose | /segformer/encoder/block.3.0/attention/self/Transpose_2 | Layer 300 - op type Transpose, Unknown input dimension, not supported by TIDL | | Split | _token_7 | Layer 331 - op type Split, Unknown input dimension, not supported by TIDL | | Transpose | Transpose_token_6 | Layer 332 - op type Transpose, Unknown input dimension, not supported by TIDL | | Squeeze | 1_squeeze_1 | Layer 335 - op type Squeeze, Unknown input dimension, not supported by TIDL | | Transpose | Transpose_token_10 | Layer 339 - op type Transpose, Unknown input dimension, not supported by TIDL | | Squeeze | 1_squeeze_0 | Subgraph does not have any compute node | | Transpose | /segformer/encoder/block.3.1/attention/self/Transpose_2 | Layer 336 - op type Transpose, Unknown input dimension, not supported by TIDL | ------------------------------------------------------------------------------------------------------------------------------------------------------- ============================= [Parsing Completed] ============================= ==================== [Optimization for subgraph_0 Started] ==================== ----------------------------- Optimization Summary ----------------------------- --------------------------------------------------------------------------------- | Layer | Nodes before optimization | Nodes after optimization | --------------------------------------------------------------------------------- | TIDL_BatchNormLayer | 0 | 12 | | TIDL_TransposeLayer | 61 | 69 | | TIDL_ConstDataLayer | 0 | 140 | | TIDL_LayerNormLayer | 26 | 26 | | TIDL_ConvolutionLayer | 16 | 12 | | TIDL_InnerProductLayer | 52 | 56 | | TIDL_EltWiseLayer | 82 | 108 | | TIDL_PoolingLayer | 2 | 2 | | TIDL_ResizeLayer | 3 | 2 | | TIDL_SoftMaxLayer | 6 | 6 | | TIDL_ErfLayer | 6 | 0 | --------------------------------------------------------------------------------- Total nodes in subgraph: 507 =================== [Optimization for subgraph_0 Completed] =================== The soft limit is 10240 The hard limit is 10240 MEM: Init ... !!! MEM: Init ... Done !!! 0.0s: VX_ZONE_INFO: Globally Enabled VX_ZONE_INFO 0.6s: VX_ZONE_INFO: Globally Enabled VX_ZONE_ERROR 0.7s: VX_ZONE_INFO: Globally Enabled VX_ZONE_WARNING 0.94s: VX_ZONE_INFO: [ownAddTargetKernelInternal:189] registered kernel vx_tutorial_graph.phase_rgb on target DSP_C7-2 0.940s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MPU-0 0.968s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MPU-1 0.990s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MPU-2 0.1003s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MPU-3 0.1022s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-1 0.1043s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-1_PRI_2 0.1058s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-1_PRI_3 0.1071s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-1_PRI_4 0.1089s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-1_PRI_5 0.1104s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-1_PRI_6 0.1115s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-1_PRI_7 0.1129s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-1_PRI_8 0.1143s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-2 0.1157s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-2_PRI_2 0.1169s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-2_PRI_3 0.1229s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-2_PRI_4 0.1244s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-2_PRI_5 0.1259s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-2_PRI_6 0.1276s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-2_PRI_7 0.1290s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-2_PRI_8 0.1309s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-3 0.1324s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-3_PRI_2 0.1341s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-3_PRI_3 0.1354s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-3_PRI_4 0.1375s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-3_PRI_5 0.1390s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-3_PRI_6 0.1406s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-3_PRI_7 0.1420s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-3_PRI_8 0.1440s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-4 0.1455s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-4_PRI_2 0.1466s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-4_PRI_3 0.1479s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-4_PRI_4 0.1491s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-4_PRI_5 0.1504s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-4_PRI_6 0.1518s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-4_PRI_7 0.1532s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-4_PRI_8 0.1547s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MCU2-0 0.1560s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target VPAC_NF 0.1572s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target VPAC_LDC1 0.1584s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target VPAC_MSC1 0.1595s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target VPAC_MSC2 0.1610s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target VPAC_VISS1 0.1622s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target CAPTURE1 0.1637s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target CAPTURE2 0.1651s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target CAPTURE3 0.1663s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target CAPTURE4 0.1676s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target CAPTURE5 0.1691s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target CAPTURE6 0.1703s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target CAPTURE7 0.1716s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target CAPTURE8 0.1729s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target CAPTURE9 0.1741s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target CAPTURE10 0.1752s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target CAPTURE11 0.1765s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target CAPTURE12 0.1781s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DISPLAY1 0.1794s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DISPLAY2 0.1805s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target CSITX 0.1820s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target CSITX2 0.1833s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSS_M2M1 0.1845s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSS_M2M2 0.1857s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSS_M2M3 0.1868s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSS_M2M4 0.1880s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target VPAC1_FC 0.1897s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MCU2-1 0.1911s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DMPAC_SDE 0.1922s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DMPAC_DOF 0.1938s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MCU3-0 0.1953s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MCU3-1 0.1967s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MCU4-0 0.1984s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target VPAC2_NF 0.1995s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target VPAC2_LDC1 0.2006s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target VPAC2_MSC1 0.2021s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target VPAC2_MSC2 0.2032s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target VPAC2_VISS1 0.2044s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target VPAC2_FC 0.2061s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MCU4-1 0.2062s: VX_ZONE_INFO: [tivxInit:152] Initialization Done !!! 0.2065s: VX_ZONE_INFO: Globally Disabled VX_ZONE_INFO ============= [Quantization & Calibration for subgraph_0 Started] ============= ==================== [Optimization for subgraph_1 Started] ==================== ----------------------------- Optimization Summary ----------------------------- --------------------------------------------------------------------------------- | Layer | Nodes before optimization | Nodes after optimization | --------------------------------------------------------------------------------- | TIDL_BatchNormLayer | 0 | 2 | | TIDL_ErfLayer | 1 | 0 | | TIDL_LayerNormLayer | 2 | 2 | | TIDL_ConstDataLayer | 0 | 12 | | TIDL_ConvolutionLayer | 1 | 1 | | TIDL_TransposeLayer | 3 | 4 | | TIDL_SqueezeLayer | 1 | 0 | | TIDL_SoftMaxLayer | 1 | 1 | | TIDL_InnerProductLayer | 6 | 6 | | TIDL_EltWiseLayer | 11 | 10 | --------------------------------------------------------------------------------- Total nodes in subgraph: 54 =================== [Optimization for subgraph_1 Completed] =================== ============= [Quantization & Calibration for subgraph_1 Started] ============= ==================== [Optimization for subgraph_2 Started] ==================== ----------------------------- Optimization Summary ----------------------------- --------------------------------------------------------------------------------- | Layer | Nodes before optimization | Nodes after optimization | --------------------------------------------------------------------------------- | TIDL_BatchNormLayer | 0 | 2 | | TIDL_EltWiseLayer | 11 | 10 | | TIDL_ConstDataLayer | 0 | 12 | | TIDL_SqueezeLayer | 1 | 0 | | TIDL_TransposeLayer | 4 | 5 | | TIDL_InnerProductLayer | 6 | 6 | | TIDL_ConvolutionLayer | 3 | 3 | | TIDL_SoftMaxLayer | 1 | 1 | | TIDL_ResizeLayer | 1 | 2 | | TIDL_LayerNormLayer | 2 | 2 | | TIDL_ConcatLayer | 1 | 1 | | TIDL_ReLULayer | 1 | 0 | | TIDL_ErfLayer | 1 | 0 | --------------------------------------------------------------------------------- Total nodes in subgraph: 64 =================== [Optimization for subgraph_2 Completed] =================== ============= [Quantization & Calibration for subgraph_2 Started] ============= -------- Running Calibration in Float Mode to Collect Tensor Statistics -------- [=============================================================================] 100 % ------------------ Fixed-point Calibration Iteration [1 / 5]: ------------------ [=============================================================================] 100 % ------------------ Fixed-point Calibration Iteration [2 / 5]: ------------------ [=============================================================================] 100 % ------------------ Fixed-point Calibration Iteration [3 / 5]: ------------------ [=============================================================================] 100 % ------------------ Fixed-point Calibration Iteration [4 / 5]: ------------------ [=============================================================================] 100 % ------------------ Fixed-point Calibration Iteration [5 / 5]: ------------------ [=============================================================================] 100 % ==================== [Quantization & Calibration Completed] ==================== ========================== [Memory Planning Started] ========================== ------------------------- Network Compiler Traces ------------------------------ Successful Memory Allocation Successful Workload Creation ========================= [Memory Planning Completed] ========================= Rerunning network compiler... ========================== [Memory Planning Started] ========================== ------------------------- Network Compiler Traces ------------------------------ Successful Memory Allocation Successful Workload Creation ========================= [Memory Planning Completed] ========================= ======================== Subgraph Compiled Successfully ======================== -------- Running Calibration in Float Mode to Collect Tensor Statistics -------- [=============================================================================] 100 % ------------------ Fixed-point Calibration Iteration [1 / 5]: ------------------ [=============================================================================] 100 % ------------------ Fixed-point Calibration Iteration [2 / 5]: ------------------ [=============================================================================] 100 % ------------------ Fixed-point Calibration Iteration [3 / 5]: ------------------ [=============================================================================] 100 % ------------------ Fixed-point Calibration Iteration [4 / 5]: ------------------ [=============================================================================] 100 % ------------------ Fixed-point Calibration Iteration [5 / 5]: ------------------ [=============================================================================] 100 % ==================== [Quantization & Calibration Completed] ==================== ========================== [Memory Planning Started] ========================== ------------------------- Network Compiler Traces ------------------------------ Successful Memory Allocation Successful Workload Creation ========================= [Memory Planning Completed] ========================= Rerunning network compiler... ========================== [Memory Planning Started] ========================== ------------------------- Network Compiler Traces ------------------------------ Successful Memory Allocation Successful Workload Creation ========================= [Memory Planning Completed] ========================= ======================== Subgraph Compiled Successfully ======================== -------- Running Calibration in Float Mode to Collect Tensor Statistics -------- [=============================================================================] 100 % ------------------ Fixed-point Calibration Iteration [1 / 5]: ------------------ [=============================================================================] 100 % ------------------ Fixed-point Calibration Iteration [2 / 5]: ------------------ [=============================================================================] 100 % ------------------ Fixed-point Calibration Iteration [3 / 5]: ------------------ [=============================================================================] 100 % ------------------ Fixed-point Calibration Iteration [4 / 5]: ------------------ [=============================================================================] 100 % ------------------ Fixed-point Calibration Iteration [5 / 5]: ------------------ [=============================================================================] 100 % ==================== [Quantization & Calibration Completed] ==================== ========================== [Memory Planning Started] ========================== ------------------------- Network Compiler Traces ------------------------------ Successful Memory Allocation Successful Workload Creation ========================= [Memory Planning Completed] ========================= Rerunning network compiler... ========================== [Memory Planning Started] ========================== ------------------------- Network Compiler Traces ------------------------------ Successful Memory Allocation Successful Workload Creation ========================= [Memory Planning Completed] ========================= ======================== Subgraph Compiled Successfully ======================== Completed_Model : 1, Name : ss-ort-nvidia-b0 , Total time : 117987.66, Offload Time : 34027.03 , DDR RW MBs : 0, Output Image File : py_out_ss-ort-nvidia-b0_ADE_val_00001801.jpg, Output Bin File : py_out_ss-ort-nvidia-b0_ADE_val_00001801.bin MEM: Deinit ... !!! MEM: Alloc's: 92 alloc's of 1110078843 bytes MEM: Free's : 92 free's of 1110078843 bytes MEM: Open's : 0 allocs of 0 bytes MEM: Deinit ... Done !!!
now i ran infer
accuracy is low on tidl and cpu
root@am69-sk:/zxseg# python3 infer_test.py
libtidl_onnxrt_EP loaded 0x148ef060
Final number of subgraphs created are : 3, - Offloaded Nodes - 382, Total Nodes - 394
APP: Init ... !!!
15821.757931 s: MEM: Init ... !!!
15821.758271 s: MEM: Initialized DMA HEAP (fd=5) !!!
15821.758576 s: MEM: Init ... Done !!!
15821.758765 s: IPC: Init ... !!!
15822.126204 s: IPC: Init ... Done !!!
REMOTE_SERVICE: Init ... !!!
REMOTE_SERVICE: Init ... Done !!!
15822.292806 s: GTC Frequency = 200 MHz
APP: Init ... Done !!!
15822.293302 s: VX_ZONE_INFO: Globally Enabled VX_ZONE_ERROR
15822.293456 s: VX_ZONE_INFO: Globally Enabled VX_ZONE_WARNING
15822.293587 s: VX_ZONE_INFO: Globally Enabled VX_ZONE_INFO
15822.294372 s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MPU-0
15822.294830 s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MPU-1
15822.295133 s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MPU-2
15822.295435 s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MPU-3
15822.295605 s: VX_ZONE_INFO: [tivxInitLocal:202] Initialization Done !!!
15822.295640 s: VX_ZONE_INFO: Globally Disabled VX_ZONE_INFO
Loaded model with providers: ['TIDLExecutionProvider', 'CPUExecutionProvider']
Output shape: (1, 150, 128, 128)
Output min/max: -44.924595 -2.3853767
Output dtype: float32
output [array([[[[ -2.7829394, -2.7829394, -2.7829394, ..., -3.1805022,
-2.7829394, -2.7829394],
[ -2.7829394, -2.7829394, -2.7829394, ..., -3.1805022,
-2.7829394, -2.7829394],
[ -3.1805022, -2.7829394, -3.1805022, ..., -3.1805022,
-3.1805022, -2.7829394],
...,
[ -3.578065 , -3.578065 , -3.578065 , ..., -3.1805022,
-3.9756277, -3.9756277],
[ -3.578065 , -3.578065 , -3.1805022, ..., -3.578065 ,
-3.578065 , -3.9756277],
[ -3.578065 , -3.578065 , -3.1805022, ..., -3.578065 ,
-3.578065 , -3.578065 ]],
[[-13.517135 , -13.517135 , -13.517135 , ..., -15.504948 ,
-15.107386 , -15.107386 ],
[-13.914697 , -13.517135 , -13.914697 , ..., -15.902511 ,
-15.504948 , -15.504948 ],
[-14.709823 , -14.31226 , -14.31226 , ..., -15.902511 ,
-15.504948 , -15.504948 ],
...,
[-15.107386 , -15.504948 , -14.709823 , ..., -11.926883 ,
-13.119572 , -13.119572 ],
[-15.107386 , -15.504948 , -14.31226 , ..., -12.324446 ,
-12.722009 , -13.119572 ],
[-14.709823 , -15.504948 , -14.31226 , ..., -13.119572 ,
-12.722009 , -13.119572 ]],
[[-11.926883 , -11.529321 , -11.529321 , ..., -14.709823 ,
-13.914697 , -13.119572 ],
[-12.324446 , -11.926883 , -11.529321 , ..., -14.709823 ,
-13.914697 , -13.914697 ],
[-13.517135 , -13.119572 , -12.722009 , ..., -14.709823 ,
-14.31226 , -13.517135 ],
...,
[-15.504948 , -16.697636 , -15.504948 , ..., -11.529321 ,
-12.324446 , -13.119572 ],
[-15.504948 , -17.0952 , -14.31226 , ..., -11.926883 ,
-11.529321 , -13.119572 ],
[-15.107386 , -15.107386 , -14.31226 , ..., -12.722009 ,
-11.529321 , -13.119572 ]],
...,
[[-21.070827 , -20.275702 , -19.87814 , ..., -22.263515 ,
-22.263515 , -22.263515 ],
[-21.46839 , -21.070827 , -20.275702 , ..., -23.058641 ,
-22.661077 , -23.058641 ],
[-22.263515 , -21.46839 , -20.673264 , ..., -23.456203 ,
-22.661077 , -23.058641 ],
...,
[-25.444017 , -26.239143 , -25.046455 , ..., -19.083014 ,
-21.46839 , -21.46839 ],
[-25.84158 , -25.84158 , -23.456203 , ..., -20.673264 ,
-21.46839 , -22.263515 ],
[-25.444017 , -27.03427 , -24.648891 , ..., -21.46839 ,
-21.46839 , -21.865952 ]],
[[-19.083014 , -18.287888 , -18.287888 , ..., -20.275702 ,
-19.87814 , -19.87814 ],
[-19.87814 , -19.083014 , -19.083014 , ..., -20.673264 ,
-20.275702 , -20.673264 ],
[-21.070827 , -19.480576 , -19.87814 , ..., -20.673264 ,
-20.275702 , -20.275702 ],
...,
[-24.648891 , -25.84158 , -23.456203 , ..., -18.68545 ,
-21.46839 , -21.865952 ],
[-24.648891 , -25.444017 , -22.263515 , ..., -20.275702 ,
-21.070827 , -22.661077 ],
[-23.456203 , -25.046455 , -23.058641 , ..., -21.865952 ,
-21.46839 , -22.263515 ]],
[[ -7.15613 , -6.7585673, -6.7585673, ..., -7.9512553,
-7.553693 , -7.553693 ],
[ -7.553693 , -6.7585673, -6.7585673, ..., -7.9512553,
-6.7585673, -7.9512553],
[ -8.348818 , -7.553693 , -7.15613 , ..., -7.9512553,
-7.15613 , -7.9512553],
...,
[-20.275702 , -21.070827 , -19.083014 , ..., -15.107386 ,
-17.0952 , -17.492762 ],
[-20.275702 , -20.673264 , -18.68545 , ..., -15.902511 ,
-16.697636 , -17.492762 ],
[-19.083014 , -20.673264 , -19.480576 , ..., -16.697636 ,
-17.0952 , -17.492762 ]]]], dtype=float32)]
Mask unique values: [ 0 18]
APP: Deinit ... !!!
REMOTE_SERVICE: Deinit ... !!!
REMOTE_SERVICE: Deinit ... Done !!!
15825.212523 s: IPC: Deinit ... !!!
15825.753588 s: IPC: DeInit ... Done !!!
15825.753657 s: MEM: Deinit ... !!!
15825.758545 s: DDR_SHARED_MEM: Alloc's: 35 alloc's of 132751508 bytes
15825.758604 s: DDR_SHARED_MEM: Free's : 35 free's of 132751508 bytes
15825.758628 s: DDR_SHARED_MEM: Open's : 0 allocs of 0 bytes
15825.758666 s: MEM: Deinit ... Done !!!
APP: Deinit ... Done !!!
root@am69-sk:/zxseg# python3 infer_test.py
EP Error 'providers' and 'provider_options' should be the same length if both are given. when using ['CPUExecutionProvider']
Falling back to ['CPUExecutionProvider'] and retrying.
Loaded model with providers: ['CPUExecutionProvider']
Output shape: (1, 150, 128, 128)
Output min/max: -37.647076 1.502726
Output dtype: float32
output [array([[[[-9.10738182e+00, -8.46180344e+00, -8.89690208e+00, ...,
-7.30843163e+00, -7.19831467e+00, -7.42499590e+00],
[-9.50431728e+00, -8.79891968e+00, -9.27894783e+00, ...,
-7.61665154e+00, -7.40585041e+00, -7.66443872e+00],
[-1.01199865e+01, -9.30077934e+00, -9.55318069e+00, ...,
-7.91833210e+00, -7.49162006e+00, -7.73509026e+00],
...,
[-8.89202118e+00, -9.60954189e+00, -8.67526245e+00, ...,
-8.45482254e+00, -9.26228619e+00, -9.42125702e+00],
[-8.94773388e+00, -9.37217522e+00, -7.73367977e+00, ...,
-8.67119598e+00, -8.89135551e+00, -9.23087692e+00],
[-8.78436470e+00, -8.72591019e+00, -7.97214413e+00, ...,
-9.18196774e+00, -8.78701019e+00, -9.19165039e+00]],
[[-1.05580688e+00, -5.19240439e-01, -6.46204770e-01, ...,
3.97472233e-02, 1.96441859e-02, -2.23712787e-01],
[-1.28475881e+00, -6.92354441e-01, -9.00263131e-01, ...,
2.34207064e-02, 8.41307044e-02, -1.85798183e-01],
[-1.68210590e+00, -9.01914299e-01, -1.04362941e+00, ...,
-1.41012087e-01, 7.51651824e-02, -1.63039312e-01],
...,
[-8.75086594e+00, -9.43372631e+00, -8.31778622e+00, ...,
-1.16131220e+01, -1.23025236e+01, -1.22214193e+01],
[-8.66691113e+00, -9.01147556e+00, -7.58560610e+00, ...,
-1.17375622e+01, -1.17695818e+01, -1.21904783e+01],
[-8.39127922e+00, -8.39451599e+00, -7.91367579e+00, ...,
-1.19990225e+01, -1.17306757e+01, -1.22769880e+01]],
[[-9.35542679e+00, -8.22606564e+00, -9.77154541e+00, ...,
-4.36007261e+00, -3.64588404e+00, -2.74775696e+00],
[-9.54254723e+00, -8.22373581e+00, -9.83346272e+00, ...,
-4.21152925e+00, -3.36076975e+00, -2.49715137e+00],
[-1.02955265e+01, -8.80494118e+00, -9.67026711e+00, ...,
-4.68067694e+00, -3.28478169e+00, -2.06064844e+00],
...,
[-1.15484982e+01, -1.39187107e+01, -1.15395317e+01, ...,
-1.06908541e+01, -1.08596764e+01, -1.14195423e+01],
[-1.20424957e+01, -1.34465389e+01, -8.34186172e+00, ...,
-1.16886873e+01, -1.00103464e+01, -1.12006750e+01],
[-1.16304359e+01, -1.15399914e+01, -7.32819128e+00, ...,
-1.30932884e+01, -9.65425110e+00, -1.13869896e+01]],
...,
[[-1.65767632e+01, -1.48137388e+01, -1.53054876e+01, ...,
-1.22598991e+01, -1.19845648e+01, -1.21397972e+01],
[-1.73164673e+01, -1.56152582e+01, -1.56489859e+01, ...,
-1.25846987e+01, -1.23050089e+01, -1.25115051e+01],
[-1.88177929e+01, -1.66239719e+01, -1.62385197e+01, ...,
-1.31708536e+01, -1.24973049e+01, -1.26569748e+01],
...,
[-1.65434132e+01, -1.66482334e+01, -1.49522409e+01, ...,
-1.73858204e+01, -1.91236134e+01, -1.93649387e+01],
[-1.65106354e+01, -1.62999249e+01, -1.37105570e+01, ...,
-1.81827908e+01, -1.84118557e+01, -1.91355877e+01],
[-1.60666618e+01, -1.56589861e+01, -1.40776854e+01, ...,
-1.88848476e+01, -1.80791683e+01, -1.91343880e+01]],
[[-1.40990629e+01, -1.18888369e+01, -1.17627220e+01, ...,
-9.66371632e+00, -9.24732304e+00, -9.20552731e+00],
[-1.45914030e+01, -1.25023279e+01, -1.20748682e+01, ...,
-9.73106861e+00, -9.32184887e+00, -9.38448811e+00],
[-1.58103552e+01, -1.32339678e+01, -1.24932852e+01, ...,
-9.88481331e+00, -9.29177380e+00, -9.32881260e+00],
...,
[-1.67702923e+01, -1.75025196e+01, -1.59494038e+01, ...,
-1.84159298e+01, -1.95513268e+01, -1.96526814e+01],
[-1.68778381e+01, -1.70773335e+01, -1.45592756e+01, ...,
-1.91143475e+01, -1.89065056e+01, -1.97410011e+01],
[-1.64891090e+01, -1.61917324e+01, -1.50849390e+01, ...,
-2.03280964e+01, -1.88051624e+01, -2.02607059e+01]],
[[-1.16451569e+01, -1.01056519e+01, -1.04362154e+01, ...,
-7.99716377e+00, -8.05412006e+00, -7.95117903e+00],
[-1.17157431e+01, -1.03467522e+01, -1.02508469e+01, ...,
-7.81100845e+00, -7.57984543e+00, -8.00262260e+00],
[-1.24734840e+01, -1.07574902e+01, -1.04642143e+01, ...,
-8.11416531e+00, -7.45732689e+00, -7.94904995e+00],
...,
[-9.75324059e+00, -1.07886009e+01, -9.89416981e+00, ...,
-1.33662148e+01, -1.47563696e+01, -1.48689194e+01],
[-9.78142452e+00, -1.02218876e+01, -8.56538582e+00, ...,
-1.35902987e+01, -1.42914600e+01, -1.45997658e+01],
[-9.69767570e+00, -9.56430054e+00, -8.45411301e+00, ...,
-1.51341333e+01, -1.42068319e+01, -1.50633011e+01]]]],
dtype=float32)]
Mask unique values: [ 1 6 11 12 20 87 149]
root@am69-sk:/zxseg# INFER SCRIPT:
import onnxruntime as ort
import numpy as np
import cv2
model = "/zxseg/ss-ort-nvidia-b0/model/optimized_model.onnx"
image_path = "/zxseg/DAT/people-walking-through-business-district-in-the-city-at-sunset_bbxoqvaod_thumbnail-1080_09.png"
artifacts_dir = "/zxseg/ss-ort-nvidia-b0/artifacts"
# artifacts_dir = "/zxseg/artifacts2"
# Set up TIDL provider options
so = ort.SessionOptions()
runtime_options = {
"artifacts_folder": artifacts_dir,
}
# providers = ['TIDLExecutionProvider', 'CPUExecutionProvider']
providers = ['CPUExecutionProvider']
provider_options = [runtime_options, {}]
ort_session = ort.InferenceSession(
model,
providers=providers,
provider_options=provider_options,
sess_options=so
)
print(f"Loaded model with providers: {ort_session.get_providers()}")
original_image = cv2.imread(image_path)
original_h, original_w = original_image.shape[:2]
image = cv2.cvtColor(original_image, cv2.COLOR_BGR2RGB)
image = cv2.resize(image, (512, 512))
image = image.astype(np.float32)
mean = np.array([123.675, 116.28, 103.53], dtype=np.float32).reshape(3, 1, 1)
scale = np.array([0.017125, 0.017507, 0.017429], dtype=np.float32).reshape(3, 1, 1)
image = np.transpose(image, (2, 0, 1)) # (3, 512, 512)
image = (image - mean) * scale
image = np.expand_dims(image, axis=0)
inputs = {ort_session.get_inputs()[0].name: image}
outputs = ort_session.run(None, inputs)
print("Output shape:", outputs[0].shape)
print("Output min/max:", outputs[0].min(), outputs[0].max())
print("Output dtype:", outputs[0].dtype)
print("output",outputs)
# Try both axis=0 and axis=-1 for argmax, print unique values
if outputs[0].shape[0] <= 20: # likely [C, H, W]
output = outputs[0][0]
mask = np.argmax(output, axis=0)
else: # possibly [1, H, W, C] or [1, C, H, W]
output = outputs[0][0] # shape: [num_classes, H, W]
mask = np.argmax(output, axis=0) # shape: [H, W]
print("Mask unique values:", np.unique(mask))
# Use mask for palette and overlay
num_classes = 150 # ADE20K has 150 classes
def get_palette(num_classes):
palette = np.random.randint(0, 255, size=(num_classes, 3), dtype=np.uint8)
palette[0] = [0, 0, 0] # background as black
return palette
palette = get_palette(num_classes)
# Resize mask to original image size
mask_resized = cv2.resize(mask.astype(np.uint8), (original_w, original_h), interpolation=cv2.INTER_NEAREST)
# Map each class index to its color
output_color = palette[mask_resized] # shape: (H, W, 3)
# Overlay the mask on the original image
overlay = cv2.addWeighted(original_image, 0.6, output_color, 0.4, 0)
cv2.imwrite("segmentation_overlay.png", overlay)
method 2 ::
ADVANCE WAY IE CUSTOM WAY
STILL SAME ERROR
wrote by using yaml and configs in model zoo
import onnxruntime as ort
import cv2
import numpy as np
import os
# Segmentation model configuration based on config.yaml
model_path = '/root/ti2/edgeai-tidl-tools/ken_seg/test_model/segformer_b0_finetuned_ade_512_512_simp.onnx'
calibration_images_path = '/root/ti2/edgeai-tidl-tools/ken_seg/DAT' # ADE20k validation images
out_dir_path = '/root/ti2/edgeai-tidl-tools/ken_seg/artifacts2'
def preprocess(image_path):
"""Preprocess image for segmentation model based on config.yaml preprocessing settings"""
img = cv2.imread(image_path)
if img is None:
raise RuntimeError(f'Failed to read image: {image_path}')
img = cv2.resize(img, (512, 512))
img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
img = img.astype(np.float32)
# Apply normalization from config.yaml
# input_mean: [123.675, 116.28, 103.53]
# input_scale: [0.017125, 0.017507, 0.017429] (which is 1/std where std=[58.395, 57.12, 57.375])
mean = np.array([123.675, 116.28, 103.53], dtype=np.float32)
scale = np.array([0.017125, 0.017507, 0.017429], dtype=np.float32)
img = (img - mean) * scale
img = img.astype(np.float32)
img = np.transpose(img, (2, 0, 1))
# Add batch dimension
img = np.expand_dims(img, axis=0)
return img
all_calib_images = [os.path.join(calibration_images_path, name) for name in os.listdir(calibration_images_path)
if name.lower().endswith(('.png', '.jpg', '.jpeg'))]
if len(all_calib_images) == 0:
raise RuntimeError(f'No calibration images found in {calibration_images_path}')
# Limit calibration images to match config (calibration_frames: 12)
max_calib_frames = 12
all_calib_images = all_calib_images[:max_calib_frames]
tidl_tools_path = os.environ.get('TIDL_TOOLS_PATH')
if not tidl_tools_path:
raise EnvironmentError('TIDL_TOOLS_PATH environment variable not set.')
os.makedirs(out_dir_path, exist_ok=True)
for root, dirs, files in os.walk(out_dir_path, topdown=False):
for f in files:
os.remove(os.path.join(root, f))
for d in dirs:
os.rmdir(os.path.join(root, d))
compile_options = {
# Core settings from runtime_options in config.yaml
'tidl_tools_path': tidl_tools_path,
'artifacts_folder': out_dir_path,
'platform': 'J7',
'version': '10.1',
'import': 'yes',
'tensor_bits': 8,
'accuracy_level': 1,
'debug_level': 0,
'inference_mode': 0,
'advanced_options:calibration_frames': len(all_calib_images),
'advanced_options:calibration_iterations': 12,
'advanced_options:quantization_scale_type': 4,
'advanced_options:activation_clipping': 1,
'advanced_options:weight_clipping': 1,
'advanced_options:bias_calibration': 1,
'advanced_options:high_resolution_optimization': 0,
'advanced_options:pre_batchnorm_fold': 1,
'advanced_options:output_feature_16bit_names_list': '',
'advanced_options:params_16bit_names_list': '',
'advanced_options:add_data_convert_ops': 3,
'ti_internal_nc_flag': 83886080,
'advanced_options:max_num_subgraph_nodes': 2048,
}
so = ort.SessionOptions()
providers = ['TIDLCompilationProvider', 'CPUExecutionProvider']
session = ort.InferenceSession(
model_path,
providers=providers,
provider_options=[compile_options, {}],
session_options=so,
)
input_name = session.get_inputs()[0].name
for img_path in all_calib_images:
img = preprocess(img_path)
session.run(None, {input_name: img})
print("DONE: Segmentation model compiled successfully with TIDL Compilation Provider.")
print(f"Artifacts saved to: {out_dir_path}")
print(f"Model input shape: {session.get_inputs()[0].shape}")
print(f"Model output shape: {session.get_outputs()[0].shape}")
print(f"Used {len(all_calib_images)} calibration images")
python3 compile_seg_b0.py ========================= [Model Compilation Started] ========================= Model compilation will perform the following stages: 1. Parsing 2. Graph Optimization 3. Quantization & Calibration 4. Memory Planning ============================== [Version Summary] ============================== ------------------------------------------------------------------------------- | TIDL Tools Version | 11_00_06_00 | ------------------------------------------------------------------------------- | C7x Firmware Version | 11_00_00_00 | ------------------------------------------------------------------------------- | Runtime Version | 1.15.0 | ------------------------------------------------------------------------------- | Model Opset Version | 17 | ------------------------------------------------------------------------------- ============================== [Parsing Started] ============================== [TIDL Import] [PARSER] WARNING: Network not identified as Object Detection network : (1) Ignore if network is not Object Detection network (2) If network is Object Detection network, please specify "model_type":"OD" as part of OSRT compilation options [TIDL Import] WARNING: Resize layer - /decode_head/Resize_3 with scales > 4 is not optimal ------------------------- Subgraph Information Summary ------------------------- ------------------------------------------------------------------------------- | Core | No. of Nodes | Number of Subgraphs | ------------------------------------------------------------------------------- | C7x | 383 | 3 | | CPU | 12 | x | ------------------------------------------------------------------------------- ------------------------------------------------------------------------------------------------------------------------------------------------------- | Node | Node Name | Reason | ------------------------------------------------------------------------------------------------------------------------------------------------------- | Split | _token_0 | Layer 295 - op type Split, Unknown input dimension, not supported by TIDL | | Transpose | Transpose | Layer 296 - op type Transpose, Unknown input dimension, not supported by TIDL | | Squeeze | 0_squeeze_1 | Layer 299 - op type Squeeze, Unknown input dimension, not supported by TIDL | | Transpose | Transpose_token_3 | Layer 303 - op type Transpose, Unknown input dimension, not supported by TIDL | | Squeeze | 0_squeeze_0 | Subgraph does not have any compute node | | Transpose | /segformer/encoder/block.3.0/attention/self/Transpose_2 | Layer 300 - op type Transpose, Unknown input dimension, not supported by TIDL | | Split | _token_7 | Layer 331 - op type Split, Unknown input dimension, not supported by TIDL | | Transpose | Transpose_token_6 | Layer 332 - op type Transpose, Unknown input dimension, not supported by TIDL | | Squeeze | 1_squeeze_1 | Layer 335 - op type Squeeze, Unknown input dimension, not supported by TIDL | | Transpose | Transpose_token_10 | Layer 339 - op type Transpose, Unknown input dimension, not supported by TIDL | | Squeeze | 1_squeeze_0 | Subgraph does not have any compute node | | Transpose | /segformer/encoder/block.3.1/attention/self/Transpose_2 | Layer 336 - op type Transpose, Unknown input dimension, not supported by TIDL | ------------------------------------------------------------------------------------------------------------------------------------------------------- ============================= [Parsing Completed] ============================= ==================== [Optimization for subgraph_0 Started] ==================== ----------------------------- Optimization Summary ----------------------------- --------------------------------------------------------------------------------- | Layer | Nodes before optimization | Nodes after optimization | --------------------------------------------------------------------------------- | TIDL_BatchNormLayer | 0 | 12 | | TIDL_TransposeLayer | 61 | 69 | | TIDL_ConstDataLayer | 0 | 67 | | TIDL_LayerNormLayer | 26 | 26 | | TIDL_ConvolutionLayer | 16 | 12 | | TIDL_InnerProductLayer | 52 | 75 | | TIDL_EltWiseLayer | 82 | 16 | | TIDL_PoolingLayer | 2 | 2 | | TIDL_ResizeLayer | 3 | 2 | | TIDL_SoftMaxLayer | 6 | 6 | | TIDL_ErfLayer | 6 | 0 | --------------------------------------------------------------------------------- Total nodes in subgraph: 361 =================== [Optimization for subgraph_0 Completed] =================== The soft limit is 10240 The hard limit is 10240 MEM: Init ... !!! MEM: Init ... Done !!! 0.0s: VX_ZONE_INFO: Globally Enabled VX_ZONE_INFO 0.14s: VX_ZONE_INFO: Globally Enabled VX_ZONE_ERROR 0.38s: VX_ZONE_INFO: Globally Enabled VX_ZONE_WARNING 0.139s: VX_ZONE_INFO: [ownAddTargetKernelInternal:189] registered kernel vx_tutorial_graph.phase_rgb on target DSP_C7-2 0.1023s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MPU-0 0.1042s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MPU-1 0.1067s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MPU-2 0.1106s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MPU-3 0.1124s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-1 0.1139s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-1_PRI_2 0.1153s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-1_PRI_3 0.1166s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-1_PRI_4 0.1183s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-1_PRI_5 0.1198s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-1_PRI_6 0.1212s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-1_PRI_7 0.1225s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-1_PRI_8 0.1237s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-2 0.1251s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-2_PRI_2 0.1262s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-2_PRI_3 0.1274s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-2_PRI_4 0.1288s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-2_PRI_5 0.1299s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-2_PRI_6 0.1312s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-2_PRI_7 0.1325s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-2_PRI_8 0.1341s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-3 0.1356s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-3_PRI_2 0.1368s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-3_PRI_3 0.1379s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-3_PRI_4 0.1395s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-3_PRI_5 0.1407s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-3_PRI_6 0.1419s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-3_PRI_7 0.1432s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-3_PRI_8 0.1448s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-4 0.1461s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-4_PRI_2 0.1472s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-4_PRI_3 0.1484s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-4_PRI_4 0.1496s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-4_PRI_5 0.1509s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-4_PRI_6 0.1521s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-4_PRI_7 0.1534s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSP_C7-4_PRI_8 0.1550s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MCU2-0 0.1563s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target VPAC_NF 0.1574s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target VPAC_LDC1 0.1585s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target VPAC_MSC1 0.1596s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target VPAC_MSC2 0.1610s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target VPAC_VISS1 0.1622s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target CAPTURE1 0.1635s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target CAPTURE2 0.1649s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target CAPTURE3 0.1661s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target CAPTURE4 0.1672s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target CAPTURE5 0.1686s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target CAPTURE6 0.1699s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target CAPTURE7 0.1712s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target CAPTURE8 0.1724s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target CAPTURE9 0.1737s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target CAPTURE10 0.1748s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target CAPTURE11 0.1759s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target CAPTURE12 0.1771s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DISPLAY1 0.1784s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DISPLAY2 0.1797s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target CSITX 0.1812s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target CSITX2 0.1829s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSS_M2M1 0.1843s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSS_M2M2 0.1857s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSS_M2M3 0.1869s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DSS_M2M4 0.1880s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target VPAC1_FC 0.1896s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MCU2-1 0.1910s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DMPAC_SDE 0.1922s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target DMPAC_DOF 0.1938s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MCU3-0 0.1951s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MCU3-1 0.1965s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MCU4-0 0.1977s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target VPAC2_NF 0.1989s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target VPAC2_LDC1 0.2001s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target VPAC2_MSC1 0.2013s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target VPAC2_MSC2 0.2026s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target VPAC2_VISS1 0.2039s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target VPAC2_FC 0.2053s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MCU4-1 0.2055s: VX_ZONE_INFO: [tivxInit:152] Initialization Done !!! 0.2058s: VX_ZONE_INFO: Globally Disabled VX_ZONE_INFO ============= [Quantization & Calibration for subgraph_0 Started] ============= ==================== [Optimization for subgraph_1 Started] ==================== ----------------------------- Optimization Summary ----------------------------- --------------------------------------------------------------------------------- | Layer | Nodes before optimization | Nodes after optimization | --------------------------------------------------------------------------------- | TIDL_BatchNormLayer | 0 | 2 | | TIDL_ErfLayer | 1 | 0 | | TIDL_LayerNormLayer | 2 | 2 | | TIDL_ConstDataLayer | 0 | 4 | | TIDL_ConvolutionLayer | 1 | 1 | | TIDL_TransposeLayer | 3 | 4 | | TIDL_SqueezeLayer | 1 | 0 | | TIDL_SoftMaxLayer | 1 | 1 | | TIDL_InnerProductLayer | 6 | 6 | | TIDL_EltWiseLayer | 11 | 2 | --------------------------------------------------------------------------------- Total nodes in subgraph: 38 =================== [Optimization for subgraph_1 Completed] =================== ============= [Quantization & Calibration for subgraph_1 Started] ============= ==================== [Optimization for subgraph_2 Started] ==================== ----------------------------- Optimization Summary ----------------------------- --------------------------------------------------------------------------------- | Layer | Nodes before optimization | Nodes after optimization | --------------------------------------------------------------------------------- | TIDL_BatchNormLayer | 0 | 2 | | TIDL_EltWiseLayer | 11 | 2 | | TIDL_ConstDataLayer | 0 | 5 | | TIDL_SqueezeLayer | 1 | 0 | | TIDL_TransposeLayer | 4 | 5 | | TIDL_InnerProductLayer | 6 | 7 | | TIDL_ConvolutionLayer | 3 | 3 | | TIDL_SoftMaxLayer | 1 | 1 | | TIDL_ResizeLayer | 1 | 2 | | TIDL_LayerNormLayer | 2 | 2 | | TIDL_ConcatLayer | 1 | 1 | | TIDL_ReLULayer | 1 | 0 | | TIDL_ErfLayer | 1 | 0 | --------------------------------------------------------------------------------- Total nodes in subgraph: 50 =================== [Optimization for subgraph_2 Completed] =================== ============= [Quantization & Calibration for subgraph_2 Started] ============= -------- Running Calibration in Float Mode to Collect Tensor Statistics -------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [1 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [2 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [3 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [4 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [5 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [6 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [7 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [8 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [9 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [10 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [11 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [12 / 12]: ----------------- [=============================================================================] 100 % ==================== [Quantization & Calibration Completed] ==================== ========================== [Memory Planning Started] ========================== ------------------------- Network Compiler Traces ------------------------------ Successful Memory Allocation Successful Workload Creation ========================= [Memory Planning Completed] ========================= Rerunning network compiler... ========================== [Memory Planning Started] ========================== ------------------------- Network Compiler Traces ------------------------------ Successful Memory Allocation Successful Workload Creation ========================= [Memory Planning Completed] ========================= ======================== Subgraph Compiled Successfully ======================== -------- Running Calibration in Float Mode to Collect Tensor Statistics -------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [1 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [2 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [3 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [4 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [5 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [6 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [7 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [8 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [9 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [10 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [11 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [12 / 12]: ----------------- [=============================================================================] 100 % ==================== [Quantization & Calibration Completed] ==================== ========================== [Memory Planning Started] ========================== ------------------------- Network Compiler Traces ------------------------------ Successful Memory Allocation Successful Workload Creation ========================= [Memory Planning Completed] ========================= Rerunning network compiler... ========================== [Memory Planning Started] ========================== ------------------------- Network Compiler Traces ------------------------------ Successful Memory Allocation Successful Workload Creation ========================= [Memory Planning Completed] ========================= ======================== Subgraph Compiled Successfully ======================== -------- Running Calibration in Float Mode to Collect Tensor Statistics -------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [1 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [2 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [3 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [4 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [5 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [6 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [7 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [8 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [9 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [10 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [11 / 12]: ----------------- [=============================================================================] 100 % ----------------- Fixed-point Calibration Iteration [12 / 12]: ----------------- [=============================================================================] 100 % ==================== [Quantization & Calibration Completed] ==================== ========================== [Memory Planning Started] ========================== ------------------------- Network Compiler Traces ------------------------------ Successful Memory Allocation Successful Workload Creation ========================= [Memory Planning Completed] ========================= Rerunning network compiler... ========================== [Memory Planning Started] ========================== ------------------------- Network Compiler Traces ------------------------------ Successful Memory Allocation Successful Workload Creation ========================= [Memory Planning Completed] ========================= ======================== Subgraph Compiled Successfully ======================== DONE: Segmentation model compiled successfully with TIDL Compilation Provider. Artifacts saved to: /root/ti2/edgeai-tidl-tools/ken_seg/artifacts2 Model input shape: [1, 3, 512, 512] Model output shape: [1, 150, 128, 128] Used 12 calibration images MEM: Deinit ... !!! MEM: Alloc's: 92 alloc's of 1501308027 bytes MEM: Free's : 92 free's of 1501308027 bytes MEM: Open's : 0 allocs of 0 bytes MEM: Deinit ... Done !!!
same issue
root@am69-sk:/zxseg# python3 infer_test.py
libtidl_onnxrt_EP loaded 0x3d72ca30
Final number of subgraphs created are : 3, - Offloaded Nodes - 383, Total Nodes - 395
APP: Init ... !!!
18178.158420 s: MEM: Init ... !!!
18178.158773 s: MEM: Initialized DMA HEAP (fd=5) !!!
18178.159101 s: MEM: Init ... Done !!!
18178.159313 s: IPC: Init ... !!!
18178.392594 s: IPC: Init ... Done !!!
REMOTE_SERVICE: Init ... !!!
REMOTE_SERVICE: Init ... Done !!!
18178.442924 s: GTC Frequency = 200 MHz
APP: Init ... Done !!!
18178.443344 s: VX_ZONE_INFO: Globally Enabled VX_ZONE_ERROR
18178.443370 s: VX_ZONE_INFO: Globally Enabled VX_ZONE_WARNING
18178.443389 s: VX_ZONE_INFO: Globally Enabled VX_ZONE_INFO
18178.444054 s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MPU-0
18178.444602 s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MPU-1
18178.445087 s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MPU-2
18178.445488 s: VX_ZONE_INFO: [tivxPlatformCreateTargetId:169] Added target MPU-3
18178.445680 s: VX_ZONE_INFO: [tivxInitLocal:202] Initialization Done !!!
18178.445721 s: VX_ZONE_INFO: Globally Disabled VX_ZONE_INFO
Loaded model with providers: ['TIDLExecutionProvider', 'CPUExecutionProvider']
Output shape: (1, 150, 128, 128)
Output min/max: -40.24002 -2.2006261
Output dtype: float32
output [array([[[[ -3.4581268, -3.4581268, -3.4581268, ..., -2.8293765,
-2.8293765, -2.8293765],
[ -3.4581268, -3.4581268, -3.4581268, ..., -3.1437516,
-2.8293765, -2.8293765],
[ -3.4581268, -3.4581268, -3.772502 , ..., -3.1437516,
-2.8293765, -2.8293765],
...,
[ -3.1437516, -3.4581268, -3.4581268, ..., -2.8293765,
-3.4581268, -3.772502 ],
[ -3.1437516, -3.4581268, -2.8293765, ..., -3.1437516,
-3.1437516, -3.4581268],
[ -3.1437516, -3.1437516, -3.1437516, ..., -3.4581268,
-3.1437516, -3.4581268]],
[[-11.946257 , -11.631881 , -11.631881 , ..., -14.775633 ,
-14.146882 , -14.146882 ],
[-12.5750065, -11.946257 , -12.260632 , ..., -15.090008 ,
-14.461258 , -14.775633 ],
[-12.889381 , -12.260632 , -12.5750065, ..., -15.404383 ,
-14.775633 , -15.090008 ],
...,
[-16.033133 , -16.661884 , -16.033133 , ..., -13.518132 ,
-14.775633 , -15.090008 ],
[-15.718758 , -16.347507 , -15.090008 , ..., -14.146882 ,
-14.146882 , -15.090008 ],
[-15.718758 , -16.033133 , -15.718758 , ..., -14.775633 ,
-14.461258 , -15.090008 ]],
[[-11.317506 , -11.003131 , -10.688755 , ..., -14.146882 ,
-13.203756 , -12.260632 ],
[-11.631881 , -11.317506 , -11.003131 , ..., -13.832507 ,
-13.203756 , -12.889381 ],
[-12.5750065, -12.260632 , -11.631881 , ..., -14.461258 ,
-13.832507 , -13.203756 ],
...,
[-14.461258 , -15.404383 , -15.090008 , ..., -11.003131 ,
-11.946257 , -12.5750065],
[-14.775633 , -16.033133 , -13.203756 , ..., -11.631881 ,
-11.003131 , -12.889381 ],
[-14.461258 , -14.461258 , -12.889381 , ..., -12.5750065,
-11.317506 , -13.518132 ]],
...,
[[-19.805635 , -18.86251 , -18.548134 , ..., -19.176886 ,
-18.548134 , -18.86251 ],
[-20.12001 , -19.49126 , -18.86251 , ..., -19.805635 ,
-19.176886 , -20.12001 ],
[-20.434385 , -19.805635 , -19.49126 , ..., -20.74876 ,
-19.49126 , -20.434385 ],
...,
[-24.206888 , -24.835638 , -23.892513 , ..., -16.347507 ,
-19.176886 , -19.49126 ],
[-23.578136 , -24.521263 , -22.006262 , ..., -17.60501 ,
-18.548134 , -19.805635 ],
[-23.263762 , -25.150013 , -23.263762 , ..., -19.176886 ,
-18.548134 , -19.49126 ]],
[[-20.12001 , -18.86251 , -19.176886 , ..., -21.37751 ,
-20.74876 , -20.74876 ],
[-20.74876 , -19.805635 , -19.805635 , ..., -21.691887 ,
-20.74876 , -21.691887 ],
[-22.006262 , -20.434385 , -20.74876 , ..., -22.320637 ,
-21.063135 , -22.006262 ],
...,
[-23.892513 , -25.464388 , -23.263762 , ..., -16.97626 ,
-19.805635 , -20.434385 ],
[-23.263762 , -24.206888 , -21.37751 , ..., -18.233759 ,
-19.176886 , -20.434385 ],
[-22.320637 , -24.206888 , -22.320637 , ..., -20.12001 ,
-19.49126 , -21.063135 ]],
[[-10.37438 , -9.431255 , -9.116879 , ..., -10.688755 ,
-10.688755 , -10.688755 ],
[-10.688755 , -9.74563 , -9.74563 , ..., -11.003131 ,
-10.060005 , -11.317506 ],
[-11.003131 , -10.060005 , -10.060005 , ..., -11.631881 ,
-10.37438 , -11.631881 ],
...,
[-22.949387 , -24.206888 , -22.320637 , ..., -15.404383 ,
-17.919384 , -18.548134 ],
[-22.949387 , -23.578136 , -21.063135 , ..., -16.347507 ,
-17.290634 , -18.233759 ],
[-22.320637 , -23.578136 , -22.006262 , ..., -17.919384 ,
-17.290634 , -18.548134 ]]]], dtype=float32)]
Mask unique values: [ 0 38 95]
APP: Deinit ... !!!
REMOTE_SERVICE: Deinit ... !!!
REMOTE_SERVICE: Deinit ... Done !!!
18181.244298 s: IPC: Deinit ... !!!
18181.732321 s: IPC: DeInit ... Done !!!
18181.732384 s: MEM: Deinit ... !!!
18181.735611 s: DDR_SHARED_MEM: Alloc's: 35 alloc's of 132765800 bytes
18181.735656 s: DDR_SHARED_MEM: Free's : 35 free's of 132765800 bytes
18181.735676 s: DDR_SHARED_MEM: Open's : 0 allocs of 0 bytes
18181.735702 s: MEM: Deinit ... Done !!!
APP: Deinit ... Done !!!
root@am69-sk:/zxseg# python3 infer_test.py
EP Error 'providers' and 'provider_options' should be the same length if both are given. when using ['CPUExecutionProvider']
Falling back to ['CPUExecutionProvider'] and retrying.
Loaded model with providers: ['CPUExecutionProvider']
Output shape: (1, 150, 128, 128)
Output min/max: -37.647076 1.502726
Output dtype: float32
output [array([[[[-9.10738182e+00, -8.46180344e+00, -8.89690208e+00, ...,
-7.30843163e+00, -7.19831467e+00, -7.42499590e+00],
[-9.50431728e+00, -8.79891968e+00, -9.27894783e+00, ...,
-7.61665154e+00, -7.40585041e+00, -7.66443872e+00],
[-1.01199865e+01, -9.30077934e+00, -9.55318069e+00, ...,
-7.91833210e+00, -7.49162006e+00, -7.73509026e+00],
...,
[-8.89202118e+00, -9.60954189e+00, -8.67526245e+00, ...,
-8.45482254e+00, -9.26228619e+00, -9.42125702e+00],
[-8.94773388e+00, -9.37217522e+00, -7.73367977e+00, ...,
-8.67119598e+00, -8.89135551e+00, -9.23087692e+00],
[-8.78436470e+00, -8.72591019e+00, -7.97214413e+00, ...,
-9.18196774e+00, -8.78701019e+00, -9.19165039e+00]],
[[-1.05580688e+00, -5.19240439e-01, -6.46204770e-01, ...,
3.97472233e-02, 1.96441859e-02, -2.23712787e-01],
[-1.28475881e+00, -6.92354441e-01, -9.00263131e-01, ...,
2.34207064e-02, 8.41307044e-02, -1.85798183e-01],
[-1.68210590e+00, -9.01914299e-01, -1.04362941e+00, ...,
-1.41012087e-01, 7.51651824e-02, -1.63039312e-01],
...,
[-8.75086594e+00, -9.43372631e+00, -8.31778622e+00, ...,
-1.16131220e+01, -1.23025236e+01, -1.22214193e+01],
[-8.66691113e+00, -9.01147556e+00, -7.58560610e+00, ...,
-1.17375622e+01, -1.17695818e+01, -1.21904783e+01],
[-8.39127922e+00, -8.39451599e+00, -7.91367579e+00, ...,
-1.19990225e+01, -1.17306757e+01, -1.22769880e+01]],
[[-9.35542679e+00, -8.22606564e+00, -9.77154541e+00, ...,
-4.36007261e+00, -3.64588404e+00, -2.74775696e+00],
[-9.54254723e+00, -8.22373581e+00, -9.83346272e+00, ...,
-4.21152925e+00, -3.36076975e+00, -2.49715137e+00],
[-1.02955265e+01, -8.80494118e+00, -9.67026711e+00, ...,
-4.68067694e+00, -3.28478169e+00, -2.06064844e+00],
...,
[-1.15484982e+01, -1.39187107e+01, -1.15395317e+01, ...,
-1.06908541e+01, -1.08596764e+01, -1.14195423e+01],
[-1.20424957e+01, -1.34465389e+01, -8.34186172e+00, ...,
-1.16886873e+01, -1.00103464e+01, -1.12006750e+01],
[-1.16304359e+01, -1.15399914e+01, -7.32819128e+00, ...,
-1.30932884e+01, -9.65425110e+00, -1.13869896e+01]],
...,
[[-1.65767632e+01, -1.48137388e+01, -1.53054876e+01, ...,
-1.22598991e+01, -1.19845648e+01, -1.21397972e+01],
[-1.73164673e+01, -1.56152582e+01, -1.56489859e+01, ...,
-1.25846987e+01, -1.23050089e+01, -1.25115051e+01],
[-1.88177929e+01, -1.66239719e+01, -1.62385197e+01, ...,
-1.31708536e+01, -1.24973049e+01, -1.26569748e+01],
...,
[-1.65434132e+01, -1.66482334e+01, -1.49522409e+01, ...,
-1.73858204e+01, -1.91236134e+01, -1.93649387e+01],
[-1.65106354e+01, -1.62999249e+01, -1.37105570e+01, ...,
-1.81827908e+01, -1.84118557e+01, -1.91355877e+01],
[-1.60666618e+01, -1.56589861e+01, -1.40776854e+01, ...,
-1.88848476e+01, -1.80791683e+01, -1.91343880e+01]],
[[-1.40990629e+01, -1.18888369e+01, -1.17627220e+01, ...,
-9.66371632e+00, -9.24732304e+00, -9.20552731e+00],
[-1.45914030e+01, -1.25023279e+01, -1.20748682e+01, ...,
-9.73106861e+00, -9.32184887e+00, -9.38448811e+00],
[-1.58103552e+01, -1.32339678e+01, -1.24932852e+01, ...,
-9.88481331e+00, -9.29177380e+00, -9.32881260e+00],
...,
[-1.67702923e+01, -1.75025196e+01, -1.59494038e+01, ...,
-1.84159298e+01, -1.95513268e+01, -1.96526814e+01],
[-1.68778381e+01, -1.70773335e+01, -1.45592756e+01, ...,
-1.91143475e+01, -1.89065056e+01, -1.97410011e+01],
[-1.64891090e+01, -1.61917324e+01, -1.50849390e+01, ...,
-2.03280964e+01, -1.88051624e+01, -2.02607059e+01]],
[[-1.16451569e+01, -1.01056519e+01, -1.04362154e+01, ...,
-7.99716377e+00, -8.05412006e+00, -7.95117903e+00],
[-1.17157431e+01, -1.03467522e+01, -1.02508469e+01, ...,
-7.81100845e+00, -7.57984543e+00, -8.00262260e+00],
[-1.24734840e+01, -1.07574902e+01, -1.04642143e+01, ...,
-8.11416531e+00, -7.45732689e+00, -7.94904995e+00],
...,
[-9.75324059e+00, -1.07886009e+01, -9.89416981e+00, ...,
-1.33662148e+01, -1.47563696e+01, -1.48689194e+01],
[-9.78142452e+00, -1.02218876e+01, -8.56538582e+00, ...,
-1.35902987e+01, -1.42914600e+01, -1.45997658e+01],
[-9.69767570e+00, -9.56430054e+00, -8.45411301e+00, ...,
-1.51341333e+01, -1.42068319e+01, -1.50633011e+01]]]],
dtype=float32)]
Mask unique values: [ 1 6 11 12 20 87 149]
root@am69-sk:/zxseg# please help,
regards Venkat
