DSH Results mAP ?

[guiyang882]

The paper in the "Deep Supervised Hashing for Fast Image Retrieval", the 12bits in CIFAR-10, I run the project and get this results.

(lp3tf4) ai-i-liuguiyang@ducserver02:~/github.com/DSH_tensorflow$ python calculate_mAP.py 
[2018-02-06 15:02:50] start!
[2018-02-06 15:02:50] read data finish
[2018-02-06 15:02:59] gt_martix finish!
[2018-02-06 15:03:08] ham_martix finish!
[2018-02-06 15:03:26] sort ham_matrix finished,start calculate mAP
[2018-02-06 15:04:38] mAP: 0.6076512586050901

The paper results is mAP: 0.6157.

[yg33717]

This is not official implementation of the paper. The result may be inconsistent with the paper.

[liuguiyangnwpu]

the batch sample hash loss is very amazing which you write.
Good Job !