Industrial Hazard Detection on Public Dataset SynSpill
84mAP@50Qwen-VL 32B + LoRA (V+L)
Evaluation results
| Method | Links | |
|---|---|---|
| Qwen-VL 32B + LoRA (V+L)Evaluation Protocol=PEFT, Training Data=Synthetic + Public Data2025.08 | 84 | |
| RF-DETREvaluation Protocol=Fine-Tuning, Training Data=Synthetic + Public Data2025.08 | 83 | |
| YOLOv11Evaluation Protocol=Fine-Tuning, Training Data=Synthetic + Public Data2025.08 | 81 | |
| Qwen-VL 7B + LoRA (V+L)Evaluation Protocol=PEFT, Training Data=Synthetic + Public Data2025.08 | 78 | |
| Qwen-VL 32BEvaluation Protocol=Zero-Shot2025.08 | 42 | |
| Qwen-VL 7BEvaluation Protocol=Zero-Shot2025.08 | 35 |