Research map / Vision-language and multimodal / Multimodal LLMs
Unification of Closed-Open Industrial Detection Scenarios: New Large-Scale Benchmarks, Challenges and Baselines
Zekai Zhang, Jinglin Zhang, Qinghui Chen, Gang Li, Da Chen, Shuainan Jing, He Wang, Dagang Li, Cong Liu, Cong Bai, Shengyong Chen
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 expert-assisted domain projection (EDP) on top of the other three components (ablation) | RTVPNet-L · MMIOC-1M open scene | AP: 15.7 -> 17.4 | Improved |
| add refined visual prompt (RVP) (ablation) | RTVPNet-L · MMIOC-1M open scene | AP: 16.5 -> 17.4 | Improved |
| add visual-text interaction (V-T) (ablation) | RTVPNet-L · MMIOC-1M open scene | AP: 15.9 -> 17.4 | Improved |
| add text-visual interaction (T-V) (ablation) | RTVPNet-L · MMIOC-1M open scene | AP: 15.6 -> 17.4 | Improved |
| add expert-assisted domain projection (ablation) | RTVPNet-S · MMIOC-1M closed scene | AP: 34.0 -> 36.7 | Improved |
| use bidirectional visual-text interaction instead of one direction (ablation) | RTVPNet-L · MMIOC-1M open scene | AP: 13.3 -> 17.4 | Improved |
| use bidirectional visual-text interaction instead of one direction (ablation) | RTVPNet-S · MMIOC-1M closed scene | AP: 31.3 -> 36.7 | Improved |
| use energy activation instead of ECA/CBAM/MCA/Self-Attn for refined visual prompts (ablation) | RTVPNet-S · MMIOC-1M open scene | AP: 14.0 -> 14.9 | Improved |
| use energy activation instead of self-attention for refined visual prompts (ablation) | RTVPNet-L · MMIOC-1M closed scene | AP: 39.9 -> 41.3 | Improved |
| use mask ratio 20% instead of 10% in domain projection (ablation) | RTVPNet-L · MMIOC-1M open scene | AP: 11.0 -> 17.4 | Improved |
| use mask ratio 20% instead of 30% (ablation) | RTVPNet-L · MMIOC-1M open scene | AP: 12.9 -> 17.4 | Improved |
| use patch size 32 instead of 128 for sparse sampling (ablation) | RTVPNet-L · MMIOC-1M open scene | AP: 13.2 -> 17.4 | Improved |
Closest papers by meaning
- Zero-Shot Learning in Industrial Scenarios: New Large-Scale Benchmark, Challenges and Baseline (AAAI 2025 · similarity 0.94)
- MMAD: A Comprehensive Benchmark for Multimodal Large Language Models in Industrial Anomaly Detection (ICLR 2025 · similarity 0.91)
- Towards Open-Vocabulary Industrial Defect Understanding with a Large-Scale Multimodal Dataset (CVPR 2026 · similarity 0.91)
- AnomalyGPT: Detecting Industrial Anomalies Using Large Vision-Language Models (AAAI 2024 · similarity 0.90)
- InCLAD: A Continual Learning Benchmark for Industrial Visual Anomaly Detection (NeurIPS 2026 · similarity 0.87)
- MVREC: A General Few-shot Defect Classification Model Using Multi-View Region-Context (AAAI 2025 · similarity 0.86)
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.
