yolov3 with changed .cfg convert

mystic123 / tensorflow-yolo-v3

Implementation of YOLO v3 object detector in Tensorflow (TF-Slim)

https://medium.com/@pawekapica_31302/implementing-yolo-v3-in-tensorflow-tf-slim-c3c55ff59dbe

Apache License 2.0

893 stars 353 forks source link

yolov3 with changed .cfg convert #96

Open TNTWEN opened 4 years ago

TNTWEN commented 4 years ago

I tried some strategies to optimize the yolov3 network structure, removing some layers and channels. Will it work if i modify yolo_v3.py according to my own network

GotG commented 3 years ago

Hello @TNTWEN

i am trying the same and modified yolo_v3.py, but the resulting frozen graph produces different detections from the original weights, and also the boxes are small. Did you manage to convert your modified yolov3.cfg successfully?

TNTWEN commented 3 years ago

Hi @GotG https://github.com/TNTWEN/OpenVINO-YOLO-Automatic-Generation In this repos，i create a tool to generate tf1.x 's code automatically for any changed yolov3/v4 model. For yolov3,you could replace mystic123\tensorflow-yolo-v3\convert_weights_pb.py with the file generated by my repos. it should work correctly!

Welcome to have a try!