Where to find the MSCOCO weights?

[stephanecharette]

I'll use this issue to track the location of the MSCOCO pre-trained .weights files.

These are the same weights you can download directly from Joseph Redmon or AlexeyAB's repos.

For the list of 80 classes used in MSCOCO, see the file cfg/coco.names which starts with "person", "bicycle", "car", etc...

[stephanecharette]

> md5sum yolov2-tiny.weights 
04904b18f0e0835d8e1d6e75d3fa93fe  yolov2-tiny.weights

https://github.com/hank-ai/darknet/releases/download/v2.0/yolov2-tiny.weights

[stephanecharette]

> md5sum yolov2.weights 
70d89ba2e180739a1c700a9ff238e354  yolov2.weights

https://github.com/hank-ai/darknet/releases/download/v2.0/yolov2.weights

[stephanecharette]

> md5sum yolov3-tiny.weights
3bcd6b390912c18924b46b26a9e7ff53  yolov3-tiny.weights

https://github.com/hank-ai/darknet/releases/download/v2.0/yolov3-tiny.weights

[stephanecharette]

> md5sum yolov3.weights
c84e5b99d0e52cd466ae710cadf6d84c  yolov3.weights

https://github.com/hank-ai/darknet/releases/download/v2.0/yolov3.weights

[stephanecharette]

> md5sum yolov4-tiny.weights
8911bf808aed305d6854c4ea48fcc731  yolov4-tiny.weights

https://github.com/hank-ai/darknet/releases/download/v2.0/yolov4-tiny.weights

[stephanecharette]

> md5sum yolov4.weights
00a4878d05f4d832ab38861b32061283  yolov4.weights

https://github.com/hank-ai/darknet/releases/download/v2.0/yolov4.weights

[stephanecharette]

> md5sum yolov7-tiny.weights
70be176996efd72c37a0e481ca169a02  yolov7-tiny.weights

https://github.com/hank-ai/darknet/releases/download/v2.0/yolov7-tiny.weights

[stephanecharette]

> md5sum yolov7.weights
028a36c8e82aa3b9c41976b37cd4b08d  yolov7.weights

https://github.com/hank-ai/darknet/releases/download/v2.0/yolov7.weights

[JfdezIkr]

One quick question: Can these weights be freely used, or are they subject to any kind of license? I am particularly interested in YOLOv7.weigths :)

[stephanecharette]

are they subject to any kind of license?

See the repo for the license. See the FAQ and look at the videos when it comes to YOLOv7: https://www.ccoderun.ca/programming/yolo_faq/#configuration_template