Do something different.
VoRA. [Fully open] [Encoder-free MLLM] Vision as LoRA
389Elysium. [ECCV 2024] Elysium: Exploring Object-level Perception in Videos via MLLM
89PTSEFormer. [ECCV 2022] PTSEFormer: Progressive Temporal-Spatial Enhanced TransFormer Towards Video Object Detection
38ByteVideoLLM. [ICCV 2025] Dynamic-VLM
28Gloma. [ICLR 2025] GLOMA
1DTS. [ICML 2026] Deep Supervision Strikes Back
1