YOLOv3 weights (Keras H5 format) convert for Tensorflow Serving

This can use to convert Tensorflow(Keras H5 format) for Tensorflow Serving

The test environment

- Python 3.7
- tensorflow 2.3.1
- numpy 1.18.5
- Pillow 8.0.1

Convert yolov3.h5 weights

Make changes to convert_tf_serving.py ,

os.environ['CUDA_VISIBLE_DEVICES'] = '0' #select the GPU 

model_path='weights.h5'  
anchors_path='main_anchors.txt'
classes_path='data_classes.txt'
score_threshold=0.5 
iou_threshold=0.25               
output_model='serving/yolov3/1'

Test Tensorflow Serving Model

os.environ['CUDA_VISIBLE_DEVICES'] = '0' #select the GPU 

input_image='1.jpg'
model_path='serving/yolov3/1'
classes_path='data_classes.txt'
out_put_image='out.jpg'
input_image_size=(416,416)

Acknowledgements

keras-yolo3 for more information.