Research map / Vision-language and multimodal / Human and embodied modeling
ProBio: A Protocol-guided Multimodal Dataset for Molecular Biology Lab
Jieming Cui, Ziren Gong, Baoxiong Jia, Siyuan Huang, Zilong Zheng, Jianzhu Ma, Yixin Zhu
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 protocol guidance (answer filtering) to YOLOv7 + StrongSORT | YOLOv7 + StrongSORT · TransST | CLS: 57.22 -> 61.94 | Improved |
| add protocol guidance to SAM + DeAOT | SAM + DeAOT · TransST | CLS: 45.83 -> 62.43 | Improved |
| add protocol guidance to SAM + DeAOT | SAM + DeAOT · TransST | PRE: 91.07 -> 92.40 | No clear effect |
| use stronger detector SAM + DeAOT instead of YOLOv7 + StrongSORT | SAM + DeAOT · TransST | PRE: 59.27 -> 91.07 | Improved |
| use SAM + DeAOT instead of YOLOv7 + StrongSORT | SAM + DeAOT · TransST | FPS: 35.8 -> 2.3 | Hurt |
| remove protocol access for human annotators | human oracle · MultiAR easy | top-1 accuracy: 98.99 -> 64.64 | Hurt |
| use SBERT instead of BERT for protocol-only | SBERT · MultiAR easy | top-1 accuracy: 56.19 -> 82.39 | Improved |
| use SBERT instead of BERT for protocol-only | SBERT · MultiAR hard | top-1 accuracy: 33.72 -> 65.03 | Improved |
| add brief protocol guidance (ActionCLIP) over best vision-only MViTv2 | ActionCLIP · MultiAR easy | top-1 accuracy: 55.28 -> 71.79 | Improved |
| add brief protocol guidance (ActionCLIP) over best vision-only MViTv2 | ActionCLIP · MultiAR hard | top-1 accuracy: 21.37 -> 37.55 | Improved |
| add SAM segmentation to ActionCLIP with detailed protocol | ActionCLIP · MultiAR hard | top-1 accuracy: 40.7 -> 46.75 | Improved |
| increase protocol granularity from brief to detailed for ActionCLIP | ActionCLIP · MultiAR hard | top-1 accuracy: 37.55 -> 40.7 | Improved |
Closest papers by meaning
- BioXArena: Benchmarking LLM Agents on Multi-Modal Biomedical Machine Learning Tasks (NeurIPS 2026 · similarity 0.83)
- MicroVQA: A Multimodal Reasoning Benchmark for Microscopy-Based Scientific Research (CVPR 2025 · similarity 0.83)
- ProAct: A Benchmark and Multimodal Framework for Structure-Aware Proactive Response (ICML 2026 · similarity 0.83)
- BioCLIP: A Vision Foundation Model for the Tree of Life (CVPR 2024 · similarity 0.81)
- ProtCLIP: Function-Informed Protein Multi-Modal Learning (AAAI 2025 · similarity 0.81)
- ScienceBoard: Evaluating Multimodal Autonomous Agents in Realistic Scientific Workflows (ICLR 2026 · similarity 0.81)
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.
