Multiple Channel Data

The LFW, CASIA FaceV5, and KZDD color datasets are used to validate the clustering ability of MKDCSC for low-quality multi-channel visual data. The LFW  and CASIA-FaceV5  datasets are the most widely used public color face datasets in the field of face recognition. They are shot under different lighting conditions and complex backgrounds, and many target objects have varying degrees of data defects. The KZDD dataset is composed of color facial images with severe occlusion obtained from the internet. Street-traffic: It is a city intersection traffic video dataset that includes movement information for the R, G, and B channels. Ultrared-traffic : It is an infrared video sequence with a resolution of  , which is a typical invisible video. Invisible video is often captured in poor light or at night, which results in low contrast and blurry images from moving objects.