Skip to content

Latest commit

 

History

History
22 lines (16 loc) · 964 Bytes

README.md

File metadata and controls

22 lines (16 loc) · 964 Bytes

Multi-Negative Deep Metric for Image Retrieval

This repository has the source code for the paper "Multi-Negative Deep Metric for Image Retrieval". For the loss layer implementation, look at here. Networks can be found here.

Example retrieval results on DeepFashion dataset.
Consumer-to-Shop

Prerequisites

  1. Tensorflow r0.9
  2. Download DeepFashion & CARS196 datasets.
  3. Download pretrained VGG16 model from here.

Training Procedure

  • Ap cluster & generate negative sampling.
  • run training scrips

Feature extraction for Test

  • extract feature
  • run test scrips

More details in the future. :)