lukqw

I have the deps in a requirements.txt which gets auto installed by sagemaker ``` --find-links https://download.openmmlab.com/mmcv/dist/cpu/torch1.10.0/index.html mmcv-full mmdet==2.24 ``` Logs of install ``` 2022-04-28T15:34:01.087+02:00 | Looking in links: https://download.openmmlab.com/mmcv/dist/cpu/torch1.10.0/index.html --...

Sadly I can not provide the output of said command. I am working with a sagemaker instance where I am not able to execute such commands AFAIK. The packages of...

from the code within this repo: ``` def representative_data_gen(): fimage = open(FLAGS.dataset).read().split() for input_value in range(10): if os.path.exists(fimage[input_value]): original_image=cv2.imread(fimage[input_value]) original_image = cv2.cvtColor(original_image, cv2.COLOR_BGR2RGB) image_data = utils.image_preprocess(np.copy(original_image), [FLAGS.input_size, FLAGS.input_size]) img_in =...

+1

@JaccoGoris did you manage to find a workaround?

I managed to solve this problem by converting the model in the following way instead: (taken from [here](https://stackoverflow.com/questions/62386682/node-number-x-reshape-failed-to-prepare-tensor-resize-with-tflite-v2-2)) ``` batch_size = 1 model = Load_Yolo_model() model.save("./yolo_model") input_shape = model.inputs[0].shape.as_list() input_shape[0]...

on a sidenote: I removed the second cv2.cvtColor from my code as well as mentioned in this issue (https://github.com/pythonlessons/TensorFlow-2.x-YOLOv3/issues/131) that's why the resized cv2 image is correctly colored

I have made the following changes to the utils functions: image_preprocess  postprocess_boxes  I do realize a PR would be easier to follow but I am kinda lazy right...

looking forward to your results!

@parthjdoshi I get around 16-17 fps with a yolov3-tiny model with relu as the activation function on device with edge tpu