search

yuanli2333 / Hadamard-Matrix-for-hashing

CVPR2020/TNNLS2023: Central Similarity Quantization/Hashing for Efficient Image and Video Retrieval
MIT License
232 stars 46 forks source link

How to create the train.txt and test.txt? #4

Open yBeOne opened 4 years ago

yBeOne commented 4 years ago

I have a question that how ro create the train.txt and test.txt the format of which are different from my own dataset . And what's the meaning of the hyper-parameters: "lambda0","lambda1","lambda2". Thank you!

yuanli2333 commented 4 years ago
  1. Hi, you can write a script to create train and test list for your own dataset, it depends on your data format.
  2. The hyper-parameters: "lambda0","lambda1","lambda2" are weights to balance the three different kind of loss: Lc, Lp and Lq.
yBeOne commented 4 years ago

The size of output is [batch,hash_bit], however, the hash_center is Hash_center[hash_label] where hash_label is equal to the non-zero of truth label. So the sizes of output and hash_center may be not equal. I can't understand this code.

Kenny-Li2023 commented 3 years ago

What does the format of train.txt and test.txt mean?

Kenny-Li2023 commented 3 years ago
  1. Hi, you can write a script to create train and test list for your own dataset, it depends on your data format.
  2. The hyper-parameters: "lambda0","lambda1","lambda2" are weights to balance the three different kind of loss: Lc, Lp and Lq.

What does the format of train.txt and test.txt mean?

Kenny-Li2023 commented 3 years ago

我有一个问题，如何ro创建火车.txt和测试.txt其格式不同于我自己的数据集。超参数的含义是什么："兰姆达0"、"兰姆达1"、"兰姆达2"。谢谢！

Hello, have you solved this problem? Can you send me a copy of the train.txt and test.text scripts you got? thank!