search

ultralytics / yolov3

YOLOv3 in PyTorch > ONNX > CoreML > TFLite
https://docs.ultralytics.com
GNU Affero General Public License v3.0
10.22k stars 3.45k forks source link

Train as one yolov3 based network for multiple dataset #1537

Closed www7890 closed 3 years ago

www7890 commented 4 years ago

❔Question

Hi, I have several well-trained yolov3 OD for my different custom dataset, such as normal traffic OD, traffic light, traffic sign, and car tail-light detection.

The main reason I trained those dataset respectively is because the object sizes are too different (Truck is huge and traffic light is tiny). But I don't think it's good idea to run so many model at the same time for real-time driving scenario.

Is there any ideas or resource to help me up build a single network which share based layers , but the rest of layers are trained for different tasks?

Thanks! Any suggestion will make great help!

glenn-jocher commented 4 years ago

@www7890 YOLOv3 and now v5 train at multiscales already. This is the default behavior, no action is required on your part other than training normally. See https://docs.ultralytics.com/yolov5/tutorials/train_custom_data

github-actions[bot] commented 3 years ago

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.