Vision CNNs trained to estimate spatial latents learned similar ventral-stream-aligned representations

Open in new window