AI Meets Beauty in ACM Multimedia 2020 3rd Place Solution

3rd place solution of the Grand Challenge of AI Meets Beauty in ACM Multimedia 2020 by team NTU-Beauty with mAP@7 0.405302

Part of the codes and pretrained models are originated from https://github.com/gniknoil/Perfect500K-Beauty-and-Personal-Care-Products-Retrieval-Challenge

Citation

@inproceedings{hou2020attention,
  title={Attention-driven Unsupervised Image Retrieval for Beauty Products with Visual and Textual Clues},
  author={Hou, Jingwen and Ji, Sijie and Wang, Annan},
  booktitle={Proceedings of the 28th ACM International Conference on Multimedia},
  year={2020}
}

Guidance

• Prepare images: https://challenge2020.perfectcorp.com/
• Environment set up: conda env create -f environment.yml (have some redundant packages)
• Prepare pretrained models: download from https://github.com/gniknoil/Perfect500K-Beauty-and-Personal-Care-Products-Retrieval-Challenge and put densenet201.t7 and seresnet152.t7 under ./pretrained
• Feature extraction: change the path to images in features.py, then python features.py. Results will be saved under ./feature.
• Search: python search_post_opt.py --test_image_path=<path_to_test_images> --result_key=<name_of_output_csv>