自动编码双瓶颈哈希：Auto-Encoding Twin-Bottleneck Hashing 模糊视频帧插值：Blurry Video Frame Interpolation Generalized ODIN 元迁移学习零样本超分：Meta-Transfer Learning for Zero-Shot Super-Resolution 3D室内场景理解：Total3DUnderstanding 从有偏训练生成无偏场景图：Unbiased Scene Graph Generation from Biased Training 视觉常识R-CNN:Visual Commonsense R-CN