There are many commonly used experimental data sets in ICCV papers. Here are some common data sets:

-**ImageNet**: It is a picture classification data set provided by Google, which contains more than 6,543.8+0,000 pictures and 654.38+0,000 categories.

-**MSCOCO**: It is a data set for image classification and object detection provided by Microsoft, which contains more than 320,000 pictures and 80 categories.

-* * cifar-10/100 * *: It is an image classification data set provided by Carnegie Mellon University, including 60,000 32x32 color pictures and 10000 gray-scale pictures.

-**PascalVOC**: It is an image classification and object detection data set provided by Oxford University, which contains more than 20 categories.