$VISN vision

Latest papers

1. SafeDriveVLA: Navigation-Conditioned World Model Dreaming for Conflict-Aware End-to-End Autonomous Driving cs.RO · 14 Oct · 1 hit

   Vision-Language-Action (VLA) models are gaining attention in autonomous driving, inheriting commonsense…

2. Representation Conditioning for Generation cs.CV · 14 Oct · 4 hits

   …methods effectively accelerate the diffusion transformer training. REPA, for example, uses a pretrained vision encoder to supervise an intermediate layer through an MLP projection head which is discarded after…

3. VITAL: From Visual Motion to Precise Action cs.CV · 14 Oct · 1 hit

   …captures the dominant maneuver but leaves centimeter-scale errors. We therefore introduce VITAL (Vision-Induced Tokens for Action Learning), an action tokenizer that couples a visually derived motion…

4. Your Transformer Is Secretly Elastic cs.LG · 14 Oct · 1 hit

   …by minimizing the model's loss under soft pruning across many sparsity targets, all at once. Across 27 vision and text transformer encoders ranging from 22M to 8B parameters, our method yields smooth degradation…

5. Graph-Reweighted Behavior Cloning with Human Interventions for Precise Robotic Manipulation cs.RO · 14 Oct · 1 hit

   Long-horizon precise manipulation produces mixed-quality data for post-training vision-language-action (VLA) models: even successful demonstrations and intervention-assisted rollouts may…

6. OLiVES: An Outdoor Low-Light Video Benchmark for Enhancement and Segmentation cs.CV · 14 Oct · 1 hit

   Low-light video processing remains an underexplored problem in computer vision, primarily due to the scarcity of high-quality ground-truth data. Existing public datasets exhibit…

7. AdaRSI: On-Demand Recursive Self-Improvement for Embodied Agents cs.CL · 14 Oct · 1 hit

   Pretrained vision-language-action models (VLAs) often fail midway through unseen long-horizon tasks. Recursive…

8. Equal CER, Unequal Damage: Severity-Weighted Character Error Rate for OCR cs.CL · 14 Oct · 1 hit

   Character Error Rate (CER), the standard metric for OCR and vision-language models (VLMs), charges a typo as much as a wrong name. We introduce semCER, which keeps CER's…