Probing the Mid-level Vision Capabilities of Self-Supervised Learning | Read Paper on Bytez