ICLR 2023 accepted papers on Video
22 of the 1,575 papers accepted at ICLR 2023 (1.4%) are on Video, compared with 0.7% at ICLR 2022. They fall into 3 topics.
See ICLR 2023 in the interactive map
Working on something in this area? Describe your idea in scime atlas to see the closest accepted papers by meaning and how crowded the spot has become.
Topics and papers
Video segmentation and tracking · 11 papers
- Token Merging: Your ViT But Faster (64 citations)
- MaskViT: Masked Visual Pre-Training for Video Prediction (44 citations)
- SlotFormer: Unsupervised Visual Dynamics Simulation with Object-Centric Models (10 citations)
- Maximizing Spatio-Temporal Entropy of Deep 3D CNNs for Efficient Video Recognition (6 citations)
- Learning Vortex Dynamics for Fluid Inference and Prediction (3 citations)
- Temporal Coherent Test Time Optimization for Robust Video Classification (3 citations)
- A Control-Centric Benchmark for Video Prediction (2 citations)
- BSTT: A Bayesian Spatial-Temporal Transformer for Sleep Staging
- Exploring perceptual straightness in learned visual representations
- MIMT: Masked Image Modeling Transformer for Video Compression
- Video Scene Graph Generation from Single-Frame Weak Supervision
Video generation · 7 papers
- Make-A-Video: Text-to-Video Generation without Text-Video Data (311 citations)
- Human Motion Diffusion Model (166 citations)
- CogVideo: Large-scale Pretraining for Text-to-Video Generation via Transformers (117 citations)
- Phenaki: Variable Length Video Generation from Open Domain Textual Descriptions (79 citations)
- Imitating Human Behaviour with Diffusion Models (22 citations)
- PV3D: A 3D Generative Model for Portrait Video Generation (3 citations)
- Towards Smooth Video Composition (2 citations)