• State
  • Locked Locked
  • Replies 7 replies
  • Subscribers 96 subscribers
  • Views 640 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

Original question:

TDA2: Running caffe-jacinto object detection using Vision SDK 3.03 TIDL_OD usecase

CCS/TDA2EG-17: TDA2 SSD Sparse model fps slow

user6065453
Intellectual 260 points

Part Number: TDA2EG-17

Tool/software: Code Composer Studio

Hi

I convert a SSD self-tranined model which detect eye , and I get 65% zero in sparse model.

When i detect from camera,i think sparse model is faster than initial.

At least, it is same as model from TI origin.

Configuration：

- 4 eve 1 dsp

- deploy.prototxt：

keep_top_k: 20
confidence_threshold: 0.15

- tidl_import_JDetNet.txt

# Default - 0
randParams = 0

# 0: Caffe, 1: TensorFlow, Default - 0
modelType = 0

# 0: Fixed quantization By tarininng Framework, 1: Dyanamic quantization by TIDL, Default - 1
quantizationStyle = 1

# quantRoundAdd/100 will be added while rounding to integer, Default - 50
quantRoundAdd = 25

numParamBits = 8
# 0 : 8bit Unsigned, 1 : 8bit Signed Default - 1
inElementType = 0

inputNetFile = "_trained_model\deploy.prototxt"
inputParamsFile = "_trained_model\ti-custom-cfg1_ssdJacintoNetV2_iter_100000.caffemodel"
outputNetFile = "_output_model\tidl_net_jdetNet_ssd.bin"
outputParamsFile = "_output_model\tidl_param_jdetNet_ssd.bin"

rawSampleInData = 0
preProcType = 4
sampleInData = "_test_image\test.jpg"
tidlStatsTool = "..\quantStatsTool\eve_test_dl_algo.out.exe"
layersGroupId = 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 1 2 1 2 1 2 1 2 1 2 1 2 1 2 1 2 1 2 1 2 1 2 2 2 2 0
conv2dKernelType = 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

Training Log：

Fullscreen traininglog.txt Download 
affe Network File : _trained_model\deploy.prototxt
Caffe Model File   : _trained_model\ti-custom-cfg1_ssdJacintoNetV2_iter_100000.caffemodel
TIDL Network File  : _output_model\tidl_net_jdetNet_ssd.bin
TIDL Model File    : _output_model\tidl_param_jdetNet_ssd.bin
Name of the Network : ssdJacintoNetV2_deploy
Num Inputs :               1
Could not find detection_out Params
 Num of Layer Detected :  57
  0, TIDL_DataLayer                , data                                      0,  -1 ,  1 ,   x ,  x ,  x ,  x ,  x ,  x ,  x ,  x ,  0 ,       0 ,       0 ,       0 ,       0 ,       1 ,       3 ,     320 ,     768 ,         0 ,
  1, TIDL_BatchNormLayer           , data/bias                                 1,   1 ,  1 ,   0 ,  x ,  x ,  x ,  x ,  x ,  x ,  x ,  1 ,       1 ,       3 ,     320 ,     768 ,       1 ,       3 ,     320 ,     768 ,    737280 ,
  2, TIDL_ConvolutionLayer         , conv1a                                    1,   1 ,  1 ,   1 ,  x ,  x ,  x ,  x ,  x ,  x ,  x ,  2 ,       1 ,       3 ,     320 ,     768 ,       1 ,      32 ,     160 ,     384 , 147456000 ,
  3, TIDL_ConvolutionLayer         , conv1b                                    1,   1 ,  1 ,   2 ,  x ,  x ,  x ,  x ,  x ,  x ,  x ,  3 ,       1 ,      32 ,     160 ,     384 ,       1 ,      32 ,      80 ,     192 , 141557760 ,
  4, TIDL_ConvolutionLayer         , res2a_branch2a                            1,   1 ,  1 ,   3 ,  x ,  x ,  x ,  x ,  x ,  x ,  x ,  4 ,       1 ,      32 ,      80 ,     192 ,       1 ,      64 ,      80 ,     192 , 283115520 ,
  5, TIDL_ConvolutionLayer         , res2a_branch2b                            1,   1 ,  1 ,   4 ,  x ,  x ,  x ,  x ,  x ,  x ,  x ,  5 ,       1 ,      64 ,      80 ,     192 ,       1 ,      64 ,      40 ,      96 , 141557760 ,
  6, TIDL_ConvolutionLayer         , res3a_branch2a                            1,   1 ,  1 ,   5 ,  x ,  x ,  x ,  x ,  x ,  x ,  x ,  6 ,       1 ,      64 ,      40 ,      96 ,       1 ,     128 ,      40 ,      96 , 283115520 ,
  7, TIDL_ConvolutionLayer         , res3a_branch2b                            1,   1 ,  1 ,   6 ,  x ,  x ,  x ,  x ,  x ,  x ,  x ,  7 ,       1 ,     128 ,      40 ,      96 ,       1 ,     128 ,      40 ,      96 , 141557760 ,
  8, TIDL_PoolingLayer             , pool3                                     1,   1 ,  1 ,   7 ,  x ,  x ,  x ,  x ,  x ,  x ,  x ,  8 ,       1 ,     128 ,      40 ,      96 ,       1 ,     128 ,      20 ,      48 ,    491520 ,
  9, TIDL_ConvolutionLayer         , res4a_branch2a                            1,   1 ,  1 ,   8 ,  x ,  x ,  x ,  x ,  x ,  x ,  x ,  9 ,       1 ,     128 ,      20 ,      48 ,       1 ,     256 ,      20 ,      48 , 283115520 ,
 10, TIDL_ConvolutionLayer         , res4a_branch2b                            1,   1 ,  1 ,   9 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 10 ,       1 ,     256 ,      20 ,      48 ,       1 ,     256 ,      10 ,      24 , 141557760 ,
 11, TIDL_ConvolutionLayer         , res5a_branch2a                            1,   1 ,  1 ,  10 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 11 ,       1 ,     256 ,      10 ,      24 ,       1 ,     512 ,      10 ,      24 , 283115520 ,
 12, TIDL_ConvolutionLayer         , res5a_branch2b                            1,   1 ,  1 ,  11 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 12 ,       1 ,     512 ,      10 ,      24 ,       1 ,     512 ,      10 ,      24 , 141557760 ,
 13, TIDL_PoolingLayer             , pool6                                     1,   1 ,  1 ,  12 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 13 ,       1 ,     512 ,      10 ,      24 ,       1 ,     512 ,       5 ,      12 ,    122880 ,
 14, TIDL_PoolingLayer             , pool7                                     1,   1 ,  1 ,  13 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 14 ,       1 ,     512 ,       5 ,      12 ,       1 ,     512 ,       3 ,       6 ,     36864 ,
 15, TIDL_PoolingLayer             , pool8                                     1,   1 ,  1 ,  14 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 15 ,       1 ,     512 ,       3 ,       6 ,       1 ,     512 ,       2 ,       3 ,     12288 ,
 16, TIDL_PoolingLayer             , pool9                                     1,   1 ,  1 ,  15 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 16 ,       1 ,     512 ,       2 ,       3 ,       1 ,     512 ,       1 ,       2 ,      4096 ,
 17, TIDL_ConvolutionLayer         , ctx_output1                               1,   1 ,  1 ,   7 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 17 ,       1 ,     128 ,      40 ,      96 ,       1 ,     256 ,      40 ,      96 , 125829120 ,
 18, TIDL_ConvolutionLayer         , ctx_output2                               1,   1 ,  1 ,  12 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 18 ,       1 ,     512 ,      10 ,      24 ,       1 ,     256 ,      10 ,      24 ,  31457280 ,
 19, TIDL_ConvolutionLayer         , ctx_output3                               1,   1 ,  1 ,  13 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 19 ,       1 ,     512 ,       5 ,      12 ,       1 ,     256 ,       5 ,      12 ,   7864320 ,
 20, TIDL_ConvolutionLayer         , ctx_output4                               1,   1 ,  1 ,  14 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 20 ,       1 ,     512 ,       3 ,       6 ,       1 ,     256 ,       3 ,       6 ,   2359296 ,
 21, TIDL_ConvolutionLayer         , ctx_output5                               1,   1 ,  1 ,  15 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 21 ,       1 ,     512 ,       2 ,       3 ,       1 ,     256 ,       2 ,       3 ,    786432 ,
 22, TIDL_ConvolutionLayer         , ctx_output6                               1,   1 ,  1 ,  16 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 22 ,       1 ,     512 ,       1 ,       2 ,       1 ,     256 ,       1 ,       2 ,    262144 ,
 23, TIDL_ConvolutionLayer         , ctx_output1/relu_mbox_loc                 1,   1 ,  1 ,  17 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 23 ,       1 ,     256 ,      40 ,      96 ,       1 ,      16 ,      40 ,      96 ,  15728640 ,
 24, TIDL_FlattenLayer             , ctx_output1/relu_mbox_loc_perm            2,   1 ,  1 ,  23 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 24 ,       1 ,      16 ,      40 ,      96 ,       1 ,       1 ,       1 ,   61440 ,         1 ,
 25, TIDL_ConvolutionLayer         , ctx_output1/relu_mbox_conf                1,   1 ,  1 ,  17 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 25 ,       1 ,     256 ,      40 ,      96 ,       1 ,      16 ,      40 ,      96 ,  15728640 ,
 26, TIDL_FlattenLayer             , ctx_output1/relu_mbox_conf_perm           2,   1 ,  1 ,  25 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 26 ,       1 ,      16 ,      40 ,      96 ,       1 ,       1 ,       1 ,   61440 ,         1 ,
 28, TIDL_ConvolutionLayer         , ctx_output2/relu_mbox_loc                 1,   1 ,  1 ,  18 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 28 ,       1 ,     256 ,      10 ,      24 ,       1 ,      24 ,      10 ,      24 ,   1474560 ,
 29, TIDL_FlattenLayer             , ctx_output2/relu_mbox_loc_perm            2,   1 ,  1 ,  28 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 29 ,       1 ,      24 ,      10 ,      24 ,       1 ,       1 ,       1 ,    5760 ,         1 ,
 30, TIDL_ConvolutionLayer         , ctx_output2/relu_mbox_conf                1,   1 ,  1 ,  18 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 30 ,       1 ,     256 ,      10 ,      24 ,       1 ,      24 ,      10 ,      24 ,   1474560 ,
 31, TIDL_FlattenLayer             , ctx_output2/relu_mbox_conf_perm           2,   1 ,  1 ,  30 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 31 ,       1 ,      24 ,      10 ,      24 ,       1 ,       1 ,       1 ,    5760 ,         1 ,
 33, TIDL_ConvolutionLayer         , ctx_output3/relu_mbox_loc                 1,   1 ,  1 ,  19 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 33 ,       1 ,     256 ,       5 ,      12 ,       1 ,      24 ,       5 ,      12 ,    368640 ,
 34, TIDL_FlattenLayer             , ctx_output3/relu_mbox_loc_perm            2,   1 ,  1 ,  33 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 34 ,       1 ,      24 ,       5 ,      12 ,       1 ,       1 ,       1 ,    1440 ,         1 ,
 35, TIDL_ConvolutionLayer         , ctx_output3/relu_mbox_conf                1,   1 ,  1 ,  19 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 35 ,       1 ,     256 ,       5 ,      12 ,       1 ,      24 ,       5 ,      12 ,    368640 ,
 36, TIDL_FlattenLayer             , ctx_output3/relu_mbox_conf_perm           2,   1 ,  1 ,  35 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 36 ,       1 ,      24 ,       5 ,      12 ,       1 ,       1 ,       1 ,    1440 ,         1 ,
 38, TIDL_ConvolutionLayer         , ctx_output4/relu_mbox_loc                 1,   1 ,  1 ,  20 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 38 ,       1 ,     256 ,       3 ,       6 ,       1 ,      24 ,       3 ,       6 ,    110592 ,
 39, TIDL_FlattenLayer             , ctx_output4/relu_mbox_loc_perm            2,   1 ,  1 ,  38 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 39 ,       1 ,      24 ,       3 ,       6 ,       1 ,       1 ,       1 ,     432 ,         1 ,
 40, TIDL_ConvolutionLayer         , ctx_output4/relu_mbox_conf                1,   1 ,  1 ,  20 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 40 ,       1 ,     256 ,       3 ,       6 ,       1 ,      24 ,       3 ,       6 ,    110592 ,
 41, TIDL_FlattenLayer             , ctx_output4/relu_mbox_conf_perm           2,   1 ,  1 ,  40 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 41 ,       1 ,      24 ,       3 ,       6 ,       1 ,       1 ,       1 ,     432 ,         1 ,
 43, TIDL_ConvolutionLayer         , ctx_output5/relu_mbox_loc                 1,   1 ,  1 ,  21 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 43 ,       1 ,     256 ,       2 ,       3 ,       1 ,      16 ,       2 ,       3 ,     24576 ,
 44, TIDL_FlattenLayer             , ctx_output5/relu_mbox_loc_perm            2,   1 ,  1 ,  43 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 44 ,       1 ,      16 ,       2 ,       3 ,       1 ,       1 ,       1 ,      96 ,         1 ,
 45, TIDL_ConvolutionLayer         , ctx_output5/relu_mbox_conf                1,   1 ,  1 ,  21 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 45 ,       1 ,     256 ,       2 ,       3 ,       1 ,      16 ,       2 ,       3 ,     24576 ,
 46, TIDL_FlattenLayer             , ctx_output5/relu_mbox_conf_perm           2,   1 ,  1 ,  45 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 46 ,       1 ,      16 ,       2 ,       3 ,       1 ,       1 ,       1 ,      96 ,         1 ,
 48, TIDL_ConvolutionLayer         , ctx_output6/relu_mbox_loc                 1,   1 ,  1 ,  22 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 48 ,       1 ,     256 ,       1 ,       2 ,       1 ,      16 ,       1 ,       2 ,      8192 ,
 49, TIDL_FlattenLayer             , ctx_output6/relu_mbox_loc_perm            2,   1 ,  1 ,  48 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 49 ,       1 ,      16 ,       1 ,       2 ,       1 ,       1 ,       1 ,      32 ,         1 ,
 50, TIDL_ConvolutionLayer         , ctx_output6/relu_mbox_conf                1,   1 ,  1 ,  22 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 50 ,       1 ,     256 ,       1 ,       2 ,       1 ,      16 ,       1 ,       2 ,      8192 ,
 51, TIDL_FlattenLayer             , ctx_output6/relu_mbox_conf_perm           2,   1 ,  1 ,  50 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 51 ,       1 ,      16 ,       1 ,       2 ,       1 ,       1 ,       1 ,      32 ,         1 ,
 53, TIDL_ConcatLayer              , mbox_loc                                  2,   6 ,  1 ,  24 , 29 , 34 , 39 , 44 , 49 ,  x ,  x , 53 ,       1 ,       1 ,       1 ,   61440 ,       1 ,       1 ,       1 ,   69200 ,         1 ,
 54, TIDL_ConcatLayer              , mbox_conf                                 2,   6 ,  1 ,  26 , 31 , 36 , 41 , 46 , 51 ,  x ,  x , 54 ,       1 ,       1 ,       1 ,   61440 ,       1 ,       1 ,       1 ,   69200 ,         1 ,
 56, TIDL_DetectionOutputLayer     , detection_out                             2,   2 ,  1 ,  53 , 54 ,  x ,  x ,  x ,  x ,  x ,  x , 56 ,       1 ,       1 ,       1 ,   69200 ,       1 ,       1 ,       1 ,     560 ,         1 ,
Total Giga Macs : 2.1931
複製了         1 個檔案。

Processing config file .\tempDir\qunat_stats_config.txt !
  0, TIDL_DataLayer                ,  0,  -1 ,  1 ,  x ,  x ,  x ,  x ,  x ,  x ,  x ,  x ,  0 ,    0 ,    0 ,
0 ,    0 ,    1 ,    3 ,  320 ,  768 ,
  1, TIDL_BatchNormLayer           ,  1,   1 ,  1 ,  0 ,  x ,  x ,  x ,  x ,  x ,  x ,  x ,  1 ,    1 ,    3 ,  320 ,  768 ,    1 ,    3 ,  320 ,  768 ,
  2, TIDL_ConvolutionLayer         ,  1,   1 ,  1 ,  1 ,  x ,  x ,  x ,  x ,  x ,  x ,  x ,  2 ,    1 ,    3 ,  320 ,  768 ,    1 ,   32 ,  160 ,  384 ,
  3, TIDL_ConvolutionLayer         ,  1,   1 ,  1 ,  2 ,  x ,  x ,  x ,  x ,  x ,  x ,  x ,  3 ,    1 ,   32 ,  160 ,  384 ,    1 ,   32 ,   80 ,  192 ,
  4, TIDL_ConvolutionLayer         ,  1,   1 ,  1 ,  3 ,  x ,  x ,  x ,  x ,  x ,  x ,  x ,  4 ,    1 ,   32 ,   80 ,  192 ,    1 ,   64 ,   80 ,  192 ,
  5, TIDL_ConvolutionLayer         ,  1,   1 ,  1 ,  4 ,  x ,  x ,  x ,  x ,  x ,  x ,  x ,  5 ,    1 ,   64 ,   80 ,  192 ,    1 ,   64 ,   40 ,   96 ,
  6, TIDL_ConvolutionLayer         ,  1,   1 ,  1 ,  5 ,  x ,  x ,  x ,  x ,  x ,  x ,  x ,  6 ,    1 ,   64 ,   40 ,   96 ,    1 ,  128 ,   40 ,   96 ,
  7, TIDL_ConvolutionLayer         ,  1,   1 ,  1 ,  6 ,  x ,  x ,  x ,  x ,  x ,  x ,  x ,  7 ,    1 ,  128 ,   40 ,   96 ,    1 ,  128 ,   40 ,   96 ,
  8, TIDL_PoolingLayer             ,  1,   1 ,  1 ,  7 ,  x ,  x ,  x ,  x ,  x ,  x ,  x ,  8 ,    1 ,  128 ,   40 ,   96 ,    1 ,  128 ,   20 ,   48 ,
  9, TIDL_ConvolutionLayer         ,  1,   1 ,  1 ,  8 ,  x ,  x ,  x ,  x ,  x ,  x ,  x ,  9 ,    1 ,  128 ,   20 ,   48 ,    1 ,  256 ,   20 ,   48 ,
 10, TIDL_ConvolutionLayer         ,  1,   1 ,  1 ,  9 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 10 ,    1 ,  256 ,   20 ,   48 ,    1 ,  256 ,   10 ,   24 ,
 11, TIDL_ConvolutionLayer         ,  1,   1 ,  1 , 10 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 11 ,    1 ,  256 ,   10 ,   24 ,    1 ,  512 ,   10 ,   24 ,
 12, TIDL_ConvolutionLayer         ,  1,   1 ,  1 , 11 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 12 ,    1 ,  512 ,   10 ,   24 ,    1 ,  512 ,   10 ,   24 ,
 13, TIDL_PoolingLayer             ,  1,   1 ,  1 , 12 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 13 ,    1 ,  512 ,   10 ,   24 ,    1 ,  512 ,    5 ,   12 ,
 14, TIDL_PoolingLayer             ,  1,   1 ,  1 , 13 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 14 ,    1 ,  512 ,
5 ,   12 ,    1 ,  512 ,    3 ,    6 ,
 15, TIDL_PoolingLayer             ,  1,   1 ,  1 , 14 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 15 ,    1 ,  512 ,
3 ,    6 ,    1 ,  512 ,    2 ,    3 ,
 16, TIDL_PoolingLayer             ,  1,   1 ,  1 , 15 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 16 ,    1 ,  512 ,
2 ,    3 ,    1 ,  512 ,    1 ,    2 ,
 17, TIDL_ConvolutionLayer         ,  1,   1 ,  1 ,  7 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 17 ,    1 ,  128 ,   40 ,   96 ,    1 ,  256 ,   40 ,   96 ,
 18, TIDL_ConvolutionLayer         ,  1,   1 ,  1 , 12 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 18 ,    1 ,  512 ,   10 ,   24 ,    1 ,  256 ,   10 ,   24 ,
 19, TIDL_ConvolutionLayer         ,  1,   1 ,  1 , 13 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 19 ,    1 ,  512 ,
5 ,   12 ,    1 ,  256 ,    5 ,   12 ,
 20, TIDL_ConvolutionLayer         ,  1,   1 ,  1 , 14 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 20 ,    1 ,  512 ,
3 ,    6 ,    1 ,  256 ,    3 ,    6 ,
 21, TIDL_ConvolutionLayer         ,  1,   1 ,  1 , 15 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 21 ,    1 ,  512 ,
2 ,    3 ,    1 ,  256 ,    2 ,    3 ,
 22, TIDL_ConvolutionLayer         ,  1,   1 ,  1 , 16 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 22 ,    1 ,  512 ,
1 ,    2 ,    1 ,  256 ,    1 ,    2 ,
 23, TIDL_ConvolutionLayer         ,  1,   1 ,  1 , 17 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 23 ,    1 ,  256 ,   40 ,   96 ,    1 ,   16 ,   40 ,   96 ,
 24, TIDL_FlattenLayer             ,  1,   1 ,  1 , 23 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 24 ,    1 ,   16 ,   40 ,   96 ,    1 ,    1 ,    1 ,61440 ,
 25, TIDL_ConvolutionLayer         ,  1,   1 ,  1 , 17 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 25 ,    1 ,  256 ,   40 ,   96 ,    1 ,   16 ,   40 ,   96 ,
 26, TIDL_FlattenLayer             ,  1,   1 ,  1 , 25 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 26 ,    1 ,   16 ,   40 ,   96 ,    1 ,    1 ,    1 ,61440 ,
 27, TIDL_ConvolutionLayer         ,  1,   1 ,  1 , 18 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 28 ,    1 ,  256 ,   10 ,   24 ,    1 ,   24 ,   10 ,   24 ,
 28, TIDL_FlattenLayer             ,  1,   1 ,  1 , 28 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 29 ,    1 ,   24 ,   10 ,   24 ,    1 ,    1 ,    1 , 5760 ,
 29, TIDL_ConvolutionLayer         ,  1,   1 ,  1 , 18 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 30 ,    1 ,  256 ,   10 ,   24 ,    1 ,   24 ,   10 ,   24 ,
 30, TIDL_FlattenLayer             ,  1,   1 ,  1 , 30 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 31 ,    1 ,   24 ,   10 ,   24 ,    1 ,    1 ,    1 , 5760 ,
 31, TIDL_ConvolutionLayer         ,  1,   1 ,  1 , 19 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 33 ,    1 ,  256 ,
5 ,   12 ,    1 ,   24 ,    5 ,   12 ,
 32, TIDL_FlattenLayer             ,  1,   1 ,  1 , 33 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 34 ,    1 ,   24 ,
5 ,   12 ,    1 ,    1 ,    1 , 1440 ,
 33, TIDL_ConvolutionLayer         ,  1,   1 ,  1 , 19 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 35 ,    1 ,  256 ,
5 ,   12 ,    1 ,   24 ,    5 ,   12 ,
 34, TIDL_FlattenLayer             ,  1,   1 ,  1 , 35 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 36 ,    1 ,   24 ,
5 ,   12 ,    1 ,    1 ,    1 , 1440 ,
 35, TIDL_ConvolutionLayer         ,  1,   1 ,  1 , 20 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 38 ,    1 ,  256 ,
3 ,    6 ,    1 ,   24 ,    3 ,    6 ,
 36, TIDL_FlattenLayer             ,  1,   1 ,  1 , 38 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 39 ,    1 ,   24 ,
3 ,    6 ,    1 ,    1 ,    1 ,  432 ,
 37, TIDL_ConvolutionLayer         ,  1,   1 ,  1 , 20 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 40 ,    1 ,  256 ,
3 ,    6 ,    1 ,   24 ,    3 ,    6 ,
 38, TIDL_FlattenLayer             ,  1,   1 ,  1 , 40 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 41 ,    1 ,   24 ,
3 ,    6 ,    1 ,    1 ,    1 ,  432 ,
 39, TIDL_ConvolutionLayer         ,  1,   1 ,  1 , 21 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 43 ,    1 ,  256 ,
2 ,    3 ,    1 ,   16 ,    2 ,    3 ,
 40, TIDL_FlattenLayer             ,  1,   1 ,  1 , 43 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 44 ,    1 ,   16 ,
2 ,    3 ,    1 ,    1 ,    1 ,   96 ,
 41, TIDL_ConvolutionLayer         ,  1,   1 ,  1 , 21 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 45 ,    1 ,  256 ,
2 ,    3 ,    1 ,   16 ,    2 ,    3 ,
 42, TIDL_FlattenLayer             ,  1,   1 ,  1 , 45 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 46 ,    1 ,   16 ,
2 ,    3 ,    1 ,    1 ,    1 ,   96 ,
 43, TIDL_ConvolutionLayer         ,  1,   1 ,  1 , 22 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 48 ,    1 ,  256 ,
1 ,    2 ,    1 ,   16 ,    1 ,    2 ,
 44, TIDL_FlattenLayer             ,  1,   1 ,  1 , 48 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 49 ,    1 ,   16 ,
1 ,    2 ,    1 ,    1 ,    1 ,   32 ,
 45, TIDL_ConvolutionLayer         ,  1,   1 ,  1 , 22 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 50 ,    1 ,  256 ,
1 ,    2 ,    1 ,   16 ,    1 ,    2 ,
 46, TIDL_FlattenLayer             ,  1,   1 ,  1 , 50 ,  x ,  x ,  x ,  x ,  x ,  x ,  x , 51 ,    1 ,   16 ,
1 ,    2 ,    1 ,    1 ,    1 ,   32 ,
 47, TIDL_ConcatLayer              ,  1,   6 ,  1 , 24 , 29 , 34 , 39 , 44 , 49 ,  x ,  x , 53 ,    1 ,    1 ,
1 ,61440 ,    1 ,    1 ,    1 ,69200 ,
 48, TIDL_ConcatLayer              ,  1,   6 ,  1 , 26 , 31 , 36 , 41 , 46 , 51 ,  x ,  x , 54 ,    1 ,    1 ,
1 ,61440 ,    1 ,    1 ,    1 ,69200 ,
 49, TIDL_DetectionOutputLayer     ,  1,   2 ,  1 , 53 , 54 ,  x ,  x ,  x ,  x ,  x ,  x , 56 ,    1 ,    1 ,
1 ,69200 ,    1 ,    1 ,    1 ,  560 ,
 50, TIDL_DataLayer                ,  0,   1 , -1 , 56 ,  x ,  x ,  x ,  x ,  x ,  x ,  x ,  0 ,    1 ,    1 ,
1 ,  560 ,    0 ,    0 ,    0 ,    0 ,
Layer ID    ,inBlkWidth  ,inBlkHeight ,inBlkPitch  ,outBlkWidth ,outBlkHeight,outBlkPitch ,numInChs    ,numOutChs
  ,numProcInChs,numLclInChs ,numLclOutChs,numProcItrs ,numAccItrs  ,numHorBlock ,numVerBlock ,inBlkChPitch,outBlkChPitc,alignOrNot
      2           72           72           72           32           32           32            3           32
         3            1            8            1            3           12            5         5184         1024
            1
      3           40           34           40           32           32           32            8            8
         8            4            8            1            2           12            5         1360         1024
            1
      4           40           22           40           32           20           32           32           64
        32            8            8            1            4            6            4          880          640
            1
      5           40           22           40           32           20           32           16           16
        16            8            8            1            2            6            4          880          640
            1
      6           40           22           40           32           20           32           64          128
        64            8            8            1            8            3            2          880          640
            1
      7           40           22           40           32           20           32           32           32
        32            8            8            1            4            3            2          880          640
            1
      9           56           22           56           48           20           48          128          256
       128            7            8            1           19            1            1         1232          960
            1
     10           56           22           56           48           20           48           64           64
        64            7            8            1           10            1            1         1232          960
            1
     11           40           12           40           32           10           32          256          512
       256            8            8            1           32            1            1          480          320
            1
     12           40           12           40           32           10           32          128          128
       128            8            8            1           16            1            1          480          320
            1
     17           32           20           32           32           20           32          128          256
       128            8            8            1           16            3            2          640          640
            1
     18           32           10           32           32           10           32          512          256
       512            8            8            1           64            1            1          320          320
            1
     19           16            5           16           16            5           16          512          256
       512            8            8            1           64            1            1           80           80
            1
     20            6            3            6            6            3            6          512          256
       512           32           32            1           16            1            1           18           18
            1
     21            3            2            3            3            2            3          512          256
       512           32           32            1           16            1            1            6            6
            1
     22            2            1            2            2            1            2          512          256
       512           32           32            1           16            1            1            2            2
            1
     23           96            4           96           96            4           96          256           16
       256           32            8            1            8            1           10          384          384
            1
     25           96            4           96           96            4           96          256           16
       256           32            8            1            8            1           10          384          384
            1
     27           24           10           24           24           10           24          256           24
       256           32           24            1            8            1            1          240          240
            1
     29           24           10           24           24           10           24          256           24
       256           32           24            1            8            1            1          240          240
            1
     31           12            5           12           12            5           12          256           24
       256           32           24            1            8            1            1           60           60
            1
     33           12            5           12           12            5           12          256           24
       256           32           24            1            8            1            1           60           60
            1
     35            6            3            6            6            3            6          256           24
       256           32           24            1            8            1            1           18           18
            1
     37            6            3            6            6            3            6          256           24
       256           32           24            1            8            1            1           18           18
            1
     39            3            2            3            3            2            3          256           16
       256           32           16            1            8            1            1            6            6
            1
     41            3            2            3            3            2            3          256           16
       256           32           16            1            8            1            1            6            6
            1
     43            2            1            2            2            1            2          256           16
       256           32           16            1            8            1            1            2            2
            1
     45            2            1            2            2            1            2          256           16
       256           32           16            1            8            1            1            2            2
            1

Processing Frame Number : 0

 Layer    1 : Out Q :      254 , TIDL_BatchNormLayer  , PASSED  #MMACs =     0.74,     0.74, Sparsity :   0.00
 Layer    2 : Out Q :     5401 , TIDL_ConvolutionLayer, PASSED  #MMACs =   147.46,    84.05, Sparsity :  43.00
 Layer    3 : Out Q :     4991 , TIDL_ConvolutionLayer, PASSED  #MMACs =   141.56,    47.19, Sparsity :  66.67
 Layer    4 : Out Q :     7339 , TIDL_ConvolutionLayer, PASSED  #MMACs =   283.12,    74.53, Sparsity :  73.68
 Layer    5 : Out Q :     4613 , TIDL_ConvolutionLayer, PASSED  #MMACs =   141.56,    42.46, Sparsity :  70.01
 Layer    6 : Out Q :     5042 , TIDL_ConvolutionLayer, PASSED  #MMACs =   283.12,    48.32, Sparsity :  82.93
 Layer    7 : Out Q :    24908 , TIDL_ConvolutionLayer, PASSED  #MMACs =   141.56,    26.85, Sparsity :  81.03
 Layer    8 :TIDL_PoolingLayer,     PASSED  #MMACs =     0.12,     0.12, Sparsity :   0.00
 Layer    9 : Out Q :     3781 , TIDL_ConvolutionLayer, PASSED  #MMACs =   283.12,     4.07, Sparsity :  98.56
 Layer   10 : Out Q :     4237 , TIDL_ConvolutionLayer, PASSED  #MMACs =   141.56,     2.05, Sparsity :  98.55
 Layer   11 : Out Q :     6494 , TIDL_ConvolutionLayer, PASSED  #MMACs =   283.12,     1.30, Sparsity :  99.54
 Layer   12 : Out Q :    16205 , TIDL_ConvolutionLayer, PASSED  #MMACs =   141.56,     0.64, Sparsity :  99.55
 Layer   13 :TIDL_PoolingLayer,     PASSED  #MMACs =     0.03,     0.03, Sparsity :   0.00
 Layer   14 :TIDL_PoolingLayer,     PASSED  #MMACs =     0.01,     0.01, Sparsity :   0.00
 Layer   15 :TIDL_PoolingLayer,     PASSED  #MMACs =     0.00,     0.00, Sparsity :   0.00
 Layer   16 :TIDL_PoolingLayer,     PASSED  #MMACs =     0.00,     0.00, Sparsity :   0.00
 Layer   17 : Out Q :    12539 , TIDL_ConvolutionLayer, PASSED  #MMACs =   125.83,    49.66, Sparsity :  60.53
 Layer   18 : Out Q :    13994 , TIDL_ConvolutionLayer, PASSED  #MMACs =    31.46,     1.03, Sparsity :  96.73
 Layer   19 : Out Q :    13587 , TIDL_ConvolutionLayer, PASSED  #MMACs =     7.86,     0.08, Sparsity :  99.04
 Layer   20 : Out Q :   121993 , TIDL_ConvolutionLayer, PASSED  #MMACs =     2.36,     2.36, Sparsity :   0.00
 Layer   21 : Out Q :  6787132 , TIDL_ConvolutionLayer, PASSED  #MMACs =     0.79,     0.79, Sparsity :   0.00
 Layer   22 : Out Q :  6812250 , TIDL_ConvolutionLayer, PASSED  #MMACs =     0.26,     0.26, Sparsity :   0.00
 Layer   23 : Out Q :     2013 , TIDL_ConvolutionLayer, PASSED  #MMACs =    15.73,    15.73, Sparsity :   0.00
 Layer   24 :TIDL_FlattenLayer, PASSED  #MMACs =     0.06,     0.06, Sparsity :   0.00
 Layer   25 : Out Q :     2326 , TIDL_ConvolutionLayer, PASSED  #MMACs =    15.73,    15.73, Sparsity :   0.00
 Layer   26 :TIDL_FlattenLayer, PASSED  #MMACs =     0.06,     0.06, Sparsity :   0.00
 Layer   27 : Out Q :     2848 , TIDL_ConvolutionLayer, PASSED  #MMACs =     1.47,     1.47, Sparsity :   0.00
 Layer   28 :TIDL_FlattenLayer, PASSED  #MMACs =     0.01,     0.01, Sparsity :   0.00
 Layer   29 : Out Q :     3505 , TIDL_ConvolutionLayer, PASSED  #MMACs =     1.47,     1.47, Sparsity :   0.00
 Layer   30 :TIDL_FlattenLayer, PASSED  #MMACs =     0.01,     0.01, Sparsity :   0.00
 Layer   31 : Out Q :     5687 , TIDL_ConvolutionLayer, PASSED  #MMACs =     0.37,     0.37, Sparsity :   0.00
 Layer   32 :TIDL_FlattenLayer, PASSED  #MMACs =     0.00,     0.00, Sparsity :   0.00
 Layer   33 : Out Q :     4071 , TIDL_ConvolutionLayer, PASSED  #MMACs =     0.37,     0.37, Sparsity :   0.00
 Layer   34 :TIDL_FlattenLayer, PASSED  #MMACs =     0.00,     0.00, Sparsity :   0.00
 Layer   35 : Out Q :    12784 , TIDL_ConvolutionLayer, PASSED  #MMACs =     0.11,     0.11, Sparsity :   0.00
 Layer   36 :TIDL_FlattenLayer, PASSED  #MMACs =     0.00,     0.00, Sparsity :   0.00
 Layer   37 : Out Q :     6509 , TIDL_ConvolutionLayer, PASSED  #MMACs =     0.11,     0.11, Sparsity :   0.00
 Layer   38 :TIDL_FlattenLayer, PASSED  #MMACs =     0.00,     0.00, Sparsity :   0.00
 Layer   39 : Out Q : 886003470 , TIDL_ConvolutionLayer, PASSED  #MMACs =     0.02,     0.02, Sparsity :   0.00
 Layer   40 :TIDL_FlattenLayer, PASSED  #MMACs =     0.00,     0.00, Sparsity :   0.00
 Layer   41 : Out Q : 949796193 , TIDL_ConvolutionLayer, PASSED  #MMACs =     0.02,     0.02, Sparsity :   0.00
 Layer   42 :TIDL_FlattenLayer, PASSED  #MMACs =     0.00,     0.00, Sparsity :   0.00
 Layer   43 : Out Q : 989614478 , TIDL_ConvolutionLayer, PASSED  #MMACs =     0.01,     0.01, Sparsity :   0.00
 Layer   44 :TIDL_FlattenLayer, PASSED  #MMACs =     0.00,     0.00, Sparsity :   0.00
 Layer   45 : Out Q : 927051167 , TIDL_ConvolutionLayer, PASSED  #MMACs =     0.01,     0.01, Sparsity :   0.00
 Layer   46 :TIDL_FlattenLayer, PASSED  #MMACs =     0.00,     0.00, Sparsity :   0.00
 Layer   47 : Out Q :     2021 , TIDL_ConcatLayer, PASSED  #MMACs =     0.00,     0.00, Sparsity :  -1.#J
 Layer   48 : Out Q :     2335 , TIDL_ConcatLayer, PASSED  #MMACs =     0.00,     0.00, Sparsity :  -1.#J
 Layer   49 : #MMACs =     0.00,     0.00, Sparsity :   0.00
End of config list found !

Result：

Log：

runlog.txt

fps 0.27x

=======================================

Q1：Are layersGroupId & conv2dKernelType correct ?

Q2：I reference link https://e2e.ti.com/support/processors/f/791/t/681674, it is not useful. why ?

Q3：anything would cause fps down when training self-model ?

Thanks,

Ting.

  • 0
    TI__Genius 17580 points

    Hi Ting,

    1. We need "deploy.prototxt" of your model to confirm layersGroupId & conv2dKernelType correct.

    2. You can refer to FAQ 21 and 22 in the TIDL user guide to correctly set and conform the layersGroupId & conv2dKernelType parameters.

    3. Yes, setting wrong parameters to layersGroupId & conv2dKernelType can cause fps down.

    Thanks,

    Praveen

  • 0 in reply to Praveen Eppa1
    Intellectual 260 points

    Hi Praveen

    1. deploy.prototxt：

    https://e2e.ti.com/cfs-file/__key/communityserver-discussions-components-files/791/1263.deploy.7z

    2. It is said "only DetectionOutputLayer should run on DSP and rest of the all the layes on EVE in the SSD network", but setting in link is different from it. why ?

    thanks for replying and helping us to check model from deploy.prototxt.

    Ting

  • 0 in reply to user6065453
    TI__Genius 17580 points

    Hi Ting,

    Yes, you can follow those settings suggested in the TIDL user guide because the settings suggested in the user guide are for the latest release. 

    The e2e link referred is based on some old release in which some of the layers are not optimized on EVE core at that time so suggested to run those on DSP core, but in the latest release all the layers are optimized on EVE core except the DetectionOutput layer. Hence in the latest release user guide it is said that "only DetectionOutputLayer should run on DSP and rest of the all the layes on EVE in the SSD network" .

    Please try with this suggestion and let us know the results.

    Thanks,

    Praveen

  • 0 in reply to Praveen Eppa1
    Intellectual 260 points

    Hi Praveen

    I follow those settings suggested in the TIDL user guide, but i still get low fps ( 0.4x - 0.6x ).

    In training step, we have three steps, i can get normal fps at initial step, like 18 - 20.

    But when i train at L1 or sparse, i will get low fps ( 0.4x - 0.6x ).

    Q：Is it should higher fps in L1 and sparse than initial ? why ?

    Thanks,

    Ting

  • 0 in reply to user6065453
    TI__Genius 17580 points

    Hi Ting,

    Could you please provide import logs in these cases to check the issue

    Thanks,

    Praveen

  • 0 in reply to Praveen Eppa1
    Intellectual 260 points

    Hi Praveen

    https://e2e.ti.com/cfs-file/__key/communityserver-discussions-components-files/791/0505_5F00_importTool.7z

    1. deploy.prototxt

    2. converting log

    3. running log

    4. import.txt

    above file, if need anything else, please tell me.

    thanks for replying

    Ting

  • 0 in reply to user6065453
    TI__Genius 17580 points

    Hi Ting,

    Thanks for sharing the logs but we did not get much required info from these logs, so could you please the entire training log (all stages) then we can look at the sparsity and get an idea.

    Regards,

    Praveen