InternImage: Exploring Large-Scale Vision Foundation Models With Deformable Convolutions | Read Paper on Bytez