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.

TDA4VM: Strided Slice

Part Number: TDA4VM

Tried to Compile Yolo v5 model from tensorflow hub on x86 machine.

Got the following error.

/root/dlrt-build/tflite_2.8/tensorflow_src/tensorflow/tensorflow/lite/kernels/strided_slice.cc:245 op_context.input->type != op_context.output->type (3 != 1)Node number 4 (STRIDED_SLICE) failed to prepare.

Tried reshaping the input images in the config file. But the error persists. 

On github, I found a collab inference of the same model using tflite intepreter, they have used stride = 32 for execution but I couldn't find which file to change in the benchmark folders. Please help

Note: I changed shaped to 640,640 which was used in the afore mentioned github repo.


'imagedet-8': dict(
task_type='detection',
calibration_dataset=imagedet_calib_dataset,
input_dataset=imagedet_val_dataset,
preprocess=preproc_transforms.get_transform_tflite((640,640), (640,640), backend='cv2'),
session=sessions.TFLiteRTSession(**tflite_session_cfg,
runtime_options=settings.runtime_options_onnx_p2(det_options=True),
model_path=f'./practice/lite-model_yolo-v5-tflite_tflite_model_1.tflite'),
postprocess=postproc_transforms.get_transform_detection_tflite(),
metric=dict(label_offset_pred=datasets.coco_det_label_offset_90to90()),
model_info=dict(metric_reference={'accuracy_ap[.5:.95]%':23.0})
)