Reading the source code of YOLOv5 object detection model
ฝัง
- เผยแพร่เมื่อ 9 ม.ค. 2024
- The content is also available as text: github.com/adensur/blog/blob/...
This video is part of my "Modern Object Detection: from YOLO to transformers" series: • Modern Object Detectio...
It goes through the process of reproducing the COCO benchmark using YOLO model, and reading the source code of the YOLOv5 implementation from ultralytics:
- Loading data and image augmentation techniques used
- Model architecture
- Loss
- Training loop
Useful links:
- My previous video about the YOLO algorithm: • YOLO object detection ...
- Notebook used in the video: github.com/adensur/blog/blob/...
- YOLOv5 quickstart guide: docs.ultralytics.com/yolov5/q...
- YOLOv5 repo: github.com/ultralytics/yolov
- COCO dataset leaderboard: paperswithcode.com/sota/objec...
- C3 convolution: arxiv.org/pdf/1812.04920.pdf
- Spatial Pyramid Pooling arxiv.org/pdf/1406.4729.pdf
- FocalLoss: arxiv.org/pdf/1708.02002v2.pdf - วิทยาศาสตร์และเทคโนโลยี
great!!
happy you liked it!
your content is gold. just matter of time before channel goes viral..
I wish you were right! Thanks for the feedback!
Thank you for the video, why not explaining yolov8 ?
I think it wasn't fully out yet when I was making the video ) But I will certainly make up for that eventually.
Although I think by now I should go for v9
thank you!!!
What other topics would be interesting for you?
@@makgaiduk Anchor free detection, Segment anything, tracking objects especially over multiple cameras. Different types of trackers like DCF DeepSort. And on the deployment side I would love to learn TensorRT. And a lot more but there are the most intersecting for me now. Keep up the good work I will be following your videos !!
thank you
You're welcome