Can I use myself dataset for training?

Hi,
The output of your deploy files are float arrays of different sizes (64, 128...). Is there a specific method you used to convert them to binary hash codes? Should I just split the values at zero, or is there some other optimal threshold?

Hi

Thank you for this wonderful paper.
Is it possible to release the training code for the algorithm ?
I wanted to use it to train for my own dataset.

Also is there any plans to release the code of this paper in tensorflow or pytorch ?

Thanks
Devraj

Hi,
Could you please include the details of how you pre-processed the image input? Specifically, the scale of image and sketch input, and whether mean subtraction was used. Thanks.

Since the dataset links you provided on github requires permission, I would like to know if you could send me the new links or permission (my email: [email protected]). I will be appreciated for your reply as soon as possible, thank you very much!

Hi, thanks for sharing the deployed model. I have read your paper Deep Sketch Hashing: Fast Free-hand Sketch-Based Image Retrieval. I'm wondering how SaN network is used for SBIR. SaN is originally designed for sketch classification and produces sketch features, but image features are not available for retrieval. Could you please explain my confusion? Thank you.

Hi, Can you please clarify about mean of sketches?
Did you use the training set to create the mean per pixel for the entire training set like alexnet or mean intensity per sketch is used