This repo is an implementation of our paper: "FashionLOGO: Prompting Multimodal Large Language Models for Fashion Logo Embeddings".
Related Datasets
- In "datasets/[dataset]" directory, we have prepared query and gallery files for each [dataset].
- Download related datasets in datasets collection, for FlickrLogos-32 dataset, you need permission from link.
Our model is released in model, download and put it into "checkpoints" folder.
python3 -m torch.distributed.launch predict.py --model_path checkpoints/your_model_checkpoint