Market-1501-C Corrupted Image Dataset

Market-1501-C is an evaluation dataset consisting of algorithmically generated variants applied to the Market-1501 test set. Portrait recognition has become a hot topic in computer vision due to its wide applications in areas such as smart security and video surveillance. However, there are still many problems in deploying current portrait recognition models to the real world. Unlike object classification and detection, portrait re-identification (ReID) is an instance-level recognition and ranking problem that relies on strong and detailed information. But current neural networks are easily corrupted by various forms of corruption, such as noise, blur, etc. This dataset contains corrupted images to help with the task of learning invariant representations for corrupted images. This dataset comprehensively establishes 5 ReID benchmarks for learning re-identification benchmarks.