Research map / Video / Video-language understanding
Go Beyond Earth: Understanding Human Actions and Scenes in Microgravity Environments
Di Wen, Lei Qi, Kunyu Peng, Kailun Yang, Fei Teng, Ao Luo, Jia Fu, Yufan Chen, Ruiping Liu, Yitian Shi, M. Saquib Sarfraz, Rainer Stiefelhagen
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 |
|---|---|---|---|
| evaluate AVA-tuned model on microgravity (MicroG-4M) instead of terrestrial JHMDB | SlowFast R101 32x2 · zero-shot, AVA fine-tuned | mAP: 47.50 -> 23.81 | Hurt |
| evaluate AVA-tuned model on microgravity (MicroG-4M) instead of terrestrial JHMDB | Slow R50 8x8 · zero-shot, AVA fine-tuned | mAP: 34.24 -> 16.24 | Hurt |
| add non-local modules to CNN backbone | I3D R50 8x8 · MicroG-4M HAR test | mAP: 46.41 -> 47.12 | Improved |
| add non-local modules to CNN backbone | C2D R50 8x8 · MicroG-4M HAR test | mAP: 29.51 -> 44.64 | Improved |
| use transformer models (MViT) instead of CNNs | MViTv1/MViTv2 · MicroG-4M HAR test | mAP: 12.86 / 15.14 vs 47.12 (I3D NLN) | Hurt |
| use longer temporal window (Slow 4x16 vs 8x8) | Slow R50 · MicroG-4M HAR validation | mAP: 40.32 -> 42.97 | Improved |
| use longer temporal window (SlowFast 4x16 vs 8x8) | SlowFast R50 · MicroG-4M HAR validation | mAP: 38.76 -> 37.10 | Hurt |
| use dense frame sampling (90 frames) for captioning | InternVideo · MicroG-4M captioning | CIDEr: 0.30 -> 0.77 (vs LLaVA-NeXT 8 frames) | Improved |
| use more input frames within the 3s window for VQA | Gemini 1.5 Pro · MicroG-4M VQA | S-VQA: 43.15 (16 frames) vs 44.75 (Gemini 2.5 Pro, 3 frames) | No clear effect |
| use closed-source instead of open-source video-language models for captioning | Gemini 1.5 Pro · MicroG-4M captioning | CIDEr: 0.77 -> 3.52 (vs InternVideo) | Improved |
Closest papers by meaning
- Sekai: A Video Dataset towards World Exploration (NeurIPS 2025 · similarity 0.83)
- FineParser: A Fine-grained Spatio-temporal Action Parser for Human-centric Action Quality Assessment (CVPR 2024 · similarity 0.83)
- Interacted Object Grounding in Spatio-Temporal Human-Object Interactions (AAAI 2025 · similarity 0.83)
- VideoGEM: Training-free Action Grounding in Videos (CVPR 2025 · similarity 0.82)
- CoSpace: Benchmarking Continuous Space Perception Ability for Vision-Language Models (CVPR 2025 · similarity 0.82)
- SpaceVista: All-Scale Visual Spatial Reasoning from mm to km (ICML 2026 · similarity 0.82)
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.
