How can I get the path ./matlab/caffe/matcaffe_batch_feat.m?

Question

How can I get the path ./matlab/caffe/matcaffe_batch_feat.m?

fanwenyuan opened this issue 7 years ago · comments

I met the 10% map error when testing the demo on cigar-10,when I was following the issue kevinlin311tw/caffe-cvprw15#17 I can't find that file.So where is it? There is only a +caffe file under my ./matlab and I can't find that file in it.

Kevin Ke-Yun Lin · Answer 1 · Sun Feb 04 2018 10:52:08 GMT+0800 (China Standard Time)

Please try the file feat_batch.m

fanwenyuan · Answer 2 · Sun Feb 04 2018 13:14:34 GMT+0800 (China Standard Time)

@kevinlin311tw Thanks.So I just need to modify that file as the FAQ referred? How can I get the KevinNet_cifar10 model and its prorotxt?

Kevin Ke-Yun Lin · Answer 3 · Mon Feb 05 2018 07:04:47 GMT+0800 (China Standard Time)

The reason you got 10% mAP is that Matlab does not load the mean file correctly. Directly specifying the path of the mean file solves this problem. For this SSDH repo, you have to use SSDH model instead of KevinNet_cifar10 model. The KevinNet_cifar10 model is for another repo. You should already downloaded the SSDH model and the corresponding prototxt if you follow the instruction.

…

On 2018年2月3日, at 下午9:14, fanwenyuan ***@***.***> wrote: @kevinlin311tw Thanks.So I just need to modify that file as the FAQ referred? How can I get the KevinNet_cifar10 model and its prorotxt? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.

fanwenyuan · Answer 4 · Mon Feb 05 2018 16:07:54 GMT+0800 (China Standard Time)

@kevinlin311tw I have specified the mean file and rebuilt the project, but still get 0.1 map.

Kevin Ke-Yun Lin · Answer 5 · Mon Feb 05 2018 16:22:15 GMT+0800 (China Standard Time)

Please also check the path of the image files. Mack sure that Matlab can read the images successfully...

fanwenyuan · Answer 6 · Mon Feb 05 2018 17:57:16 GMT+0800 (China Standard Time)

Thanks. The problem has solved. After specify the path, I deleted the binary-train.mat and binary-test.mat under /analysis and reproduce them. Or the code will load the error ones directly and skip the feat_batch.m. Hope it can be helpful.