UltraGlobal

This is the repository of paper : "UltraGlobal: An Enhanced Approach to Image Retrieval Using Global Features"

Install

• You may refer to our requirement to set up the same environment with our original code.

  requirements.txt

Set up

• You can refer to ShihaoShao-GH/SuperGlobal: ICCV 2023 Paper Global Features are All You Need for Image Retrieval and Reranking Official Repository to set up the respository.

Evaluation

• You can run the instruction below to evaluate the model:

   python test.py MODEL.DEPTH 50 TEST.WEIGHTS ./weights TEST.DATA_DIR ./revisitop SupG.gemp True SupG.rgem True SupG.sgem True SupG.relup True SupG.rerank True SupG.onemeval False
  

Toxic finetuning

• You should run new_try.py under folder finetune to do toxic finetuning.

• The model is saved under ./finetune/model run the following to evaluate the toxic model:

  python test.py MODEL.DEPTH 50 TEST.WEIGHTS ./finetune/model TEST.DATA_DIR ./revisitop SupG.gemp True SupG.rgem True SupG.sgem True SupG.relup True SupG.rerank True SupG.onemeval False