search

ziqi-jin / finetune-anything

Fine-tune SAM (Segment Anything Model) for computer vision tasks such as semantic segmentation, matting, detection ... in specific scenarios
MIT License
766 stars 55 forks source link

mIoU #35

Closed eternal-f1ame closed 1 year ago

eternal-f1ame commented 1 year ago

I would like to know, from anyone who has used the available script for a semantic seg task. What was the best mIoU you could obtain. And using how many epochs/batch size.

ziqi-jin commented 1 year ago

For the voc dataset of the semantic segmentation task, bs is set to 8, which requires tens of thousands of iterations. mIOU can reach 79. I have not adjusted the parameters on purpose. The more iterations, the better the effect. epoch = iterations * batchsize / photonumber