Research map / Video / Video-language understanding
Describe Anything Anywhere At Any Moment
Nicolas Gorlo, Lukas Schmid, Luca Carlone
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add DAM descriptions instead of image crops for the agent (ablation) | DAAAM · OC-NaVQA positional error | positional error (m): 50.05 -> 41.75 | Improved |
| add DAM descriptions instead of image crops for the agent (ablation) | DAAAM · OC-NaVQA temporal error | temporal error (min): 2.396 -> 1.792 | Improved |
| add DAM descriptions instead of image crops for the agent (ablation) | DAAAM · OC-NaVQA question accuracy | question accuracy: 0.776 -> 0.711 | Hurt |
| add hierarchical region clustering (ablation) | DAAAM · OC-NaVQA temporal error | temporal error (min): 3.576 -> 1.792 | Improved |
| add hierarchical region clustering (ablation) | DAAAM · OC-NaVQA question accuracy | question accuracy: 0.707 -> 0.711 | Improved |
| add frame selection quality heuristic (ablation) | DAAAM · OC-NaVQA question accuracy | question accuracy: 0.627 -> 0.711 | Improved |
| add frame selection quality heuristic (ablation) | DAAAM · OC-NaVQA positional error | positional error (m): 49.92 -> 41.75 | Improved |
| add frame selection quality heuristic (ablation) | DAAAM · OC-NaVQA temporal error | temporal error (min): 1.678 -> 1.792 | No clear effect |
| concatenate DAM sentence embedding with CLIP feature (ablation) | CLIP ViT-L/14 retrieval · refCOCOg Top-1 | retrieval accuracy (%): 19.59 -> 25.11 | Improved |
| concatenate DAM sentence embedding with CLIP feature (ablation) | CLIP ViT-L/14 retrieval · Visual Genome Top-1 | retrieval accuracy (%): 21.64 -> 24.92 | Improved |
| use DAM description embeddings alone instead of CLIP for retrieval (ablation) | retrieval · refCOCOg Top-1 | retrieval accuracy (%): 19.59 -> 18.07 | Hurt |
| batch DAM inference (batch size 48-128) with frame selection | DAM inference · time per mask | speedup: order of magnitude | Improved |
Closest papers by meaning
- R4: Retrieval-Augmented Reasoning for Vision-Language Models in 4D Spatio-Temporal Space (CVPR 2026 · similarity 0.87)
- OST-Bench: Evaluating the Capabilities of MLLMs in Online Spatio-temporal Scene Understanding (NeurIPS 2025 · similarity 0.87)
- Stitch and Tell: A Structured Data Augmentation Method for Spatial Understanding (NeurIPS 2025 · similarity 0.87)
- Perceive Anything: Recognize, Explain, Caption, and Segment Anything in Images and Videos (NeurIPS 2025 · similarity 0.87)
- PanoWorld: Towards Spatial Supersensing in 360◦ Panorama World (NeurIPS 2026 · similarity 0.87)
- Learning to Reason in 4D: Dynamic Spatial Understanding for Vision Language Models (CVPR 2026 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
