Early end

[lccui]

❔Question

python train.py --data data/custom.data --weights pretrained/yolov4.conv.137 --batch-size 8 --cfg cfg/yolov4.cfg --img-size 256 Namespace(adam=False, batch_size=8, bucket='', cache_images=False, cfg='cfg/ship-yolov4.cfg', data='data/4wship.data', device='', epochs=300, evolve=False, freeze_layers=False, img_size=[256], multi_scale=False, name='', nosave=False, notest=False, rect=False, resume=False, single_cls=False, weights='pretrained/yolov4.conv.137')

Additional context

run：python train.py --data data/custom.data --weights pretrained/yolov3-spp.pt --batch-size 8 --cfg cfg/yolov3-spp.cfg --img-size 256 is OK，normal

[github-actions[bot]]

Hello @lccui, thank you for your interest in our work! Ultralytics has open-sourced YOLOv5 at https://github.com/ultralytics/yolov5, featuring faster, lighter and more accurate object detection. YOLOv5 is recommended for all new projects.

To continue with this repo, please visit our Custom Training Tutorial to get started, and see our Google Colab Notebook, Docker Image, and GCP Quickstart Guide for example environments.

If this is a bug report, please provide screenshots and minimum viable code to reproduce your issue, otherwise we can not help you.

If this is a custom model or data training question, please note that Ultralytics does not provide free personal support. As a leader in vision ML and AI, we do offer professional consulting, from simple expert advice up to delivery of fully customized, end-to-end production solutions for our clients, such as:

• Cloud-based AI systems operating on hundreds of HD video streams in realtime.
• Edge AI integrated into custom iOS and Android apps for realtime 30 FPS video inference.
• Custom data training, hyperparameter evolution, and model exportation to any destination.

For more information please visit https://www.ultralytics.com.

[lccui]

I already use open-sourced YOLOv5，but I still want to know why

[glenn-jocher]

@lccui strongly recommend moving from yolov3 to yolov5, as training instabilities have been fixed there. See https://github.com/ultralytics/yolov5/issues/471#issuecomment-662009779

[glenn-jocher]

I'll reference my YOLOv5 comments here for the benefit of all the local YOLOv3 watchers.

The original yolo/darknet box equations have a serious flaw. Width and Height are completely unbounded as they are simply out=exp(in), which is dangerous, as it can lead to runaway gradients, instabilities, NaN losses and ultimately a complete loss of training. yolov3 suffers from this problem as well as yolov4.

For yolov5 I made sure to patch this error by sigmoiding all model outputs, while also ensuring that the centerpoint remained unchanged 1=fcn(0), so nominal zero outputs from the model would cause the nominal anchor size to be used. The current eqn constrains anchor multiples from a minimum of 0 to a maximum of 4, and the anchor-target matching has also been updated to be width-height multiple based, with a nominal upper threshold hyperparameter of 4.0.

The original thread is https://github.com/ultralytics/yolov3/issues/168

Originally posted by @glenn-jocher in https://github.com/ultralytics/yolov5/issues/471#issuecomment-662009779

[github-actions[bot]]

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.