https://arxiv.org/abs/2012.00364
Pre-Trained Image Processing Transformer (Hanting Chen, Yunhe Wang, Tianyu Guo, Chang Xu, Yiping Deng, Zhenhua Liu, Siwei Ma, Chunjing Xu, Chao Xu, Wen Gao)
트랜스포머를 이미지 denoising, deraining, superresolution, contrastive learning으로 프리트레이닝한 다음 파인튜닝. 이건 대체...뭔가...뭔가 일어나고 있음.
#contrastive_learning #pretraining #vision_transformer #restoration