search

YuHengsss / YOLOV

This repo is an implementation of PyTorch version YOLOV Series
Apache License 2.0
278 stars 39 forks source link

training for coco form #55

Open josh3255 opened 1 year ago

josh3255 commented 1 year ago

Dear authorsreally impressive approach and great results! Thank you for publishing your work.

I don't have data in VID format, but only data in coco format, is it impossible to train the dataset in image format?

YuHengsss commented 1 year ago

Hey, you can train this video objector as a general object detetor. But you have to change the code of the data loading. Previous works [e.g. Relation Networks for Object Detection] show the relation module is also benefical to object detection in single frame, but in this work, we don't test its performance.

Rita-cw commented 3 months ago

Hey, you can train this video objector as a general object detetor. But you have to change the code of the data loading. Previous works [e.g. Relation Networks for Object Detection] show the relation module is also benefical to object detection in single frame, but in this work, we don't test its performance.

Dear author, thanks for your sharing! I don't need training but target detection based on TUM datasets(Continuous RGB images). How to run the detection task based on the TUM datasets? which file should I change the code of the data loading?Could you be more specific? thank you very much!!