UniNet-Pytorch icon indicating copy to clipboard operation
UniNet-Pytorch copied to clipboard

Iris segmentation

Open ZuoyanL opened this issue 4 years ago • 4 comments

I noticed that the results after iris segmentation were inaccuracy. Could you please tell me how can I get the excellent segmentation results with the scripts.

ZuoyanL avatar Sep 22 '20 08:09 ZuoyanL

You may need to finetune this model on your dataset. Unfortunately, I just reproduced this model and converted the parameters from caffe to pytorch, and the author did not release the training code.

Debatrix avatar Sep 28 '20 14:09 Debatrix

You may need to finetune this model on your dataset. Unfortunately, I just reproduced this model and converted the parameters from caffe to pytorch, and the author did not release the training code.

Do you know how to get the accuracy IRIS segmentation results? You know, I can't get the pupils using the scripts from either yours or the author's.

ZuoyanL avatar Sep 28 '20 14:09 ZuoyanL

Do you know how to get the accuracy IRIS segmentation results? You know, I can't get the pupils using the scripts from either yours or the author's.

The most accurate method is to segment manually. 😀 As far as I know, there are relatively few open source codes in this field, you can try Osiris, TVM, IrisParseNet and IrisSegBenchmark

Debatrix avatar Sep 28 '20 14:09 Debatrix

Do you know how to get the accuracy IRIS segmentation results? You know, I can't get the pupils using the scripts from either yours or the author's.

The most accurate method is to segment manually. 😀 As far as I know, there are relatively few open source codes in this field, you can try Osiris, TVM, IrisParseNet and IrisSegBenchmark

Yeah, I see. Thank you very much. by the way, do you use wechat?

ZuoyanL avatar Sep 29 '20 06:09 ZuoyanL