Name Extraction on BPMN Diagrams (test)
83PrecisionMistral-small-3.1
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Mistral-small-3.1Pipeline=Only VLM2025.11 | 83 | 78.1 | 80.2 | |
| GPT-4.1Pipeline=Only VLM2025.11 | 82.9 | 79.4 | 80.7 | |
| GPT-4.1-miniPipeline=Only VLM2025.11 | 82.3 | 81.9 | 81.7 | |
| GPT-4oPipeline=Only VLM2025.11 | 81.2 | 77.5 | 79 | |
| Gemma-12BPipeline=Only VLM2025.11 | 79.8 | 56 | 64.5 | |
| Mistral-small-3.1Pipeline=VLM + Tesseract2025.11 | 78.7 | 77.4 | 77.4 | |
| Sketch2ProcessPipeline=Non-VLM Baseline2025.11 | 78.7 | 63 | 69.3 | |
| GPT-4.1Pipeline=VLM + Pix2Struct2025.11 | 78.3 | 81 | 79.2 | |
| GPT-4o-miniPipeline=Only VLM2025.11 | 78.2 | 62.9 | 69.1 | |
| GPT-4.1Pipeline=VLM + RapidOCR2025.11 | 78.1 | 80.2 | 78.7 | |
| GPT-4.1Pipeline=VLM + Tesseract2025.11 | 77.7 | 78.6 | 77.5 | |
| Mistral-small-3.1Pipeline=VLM + Pix2Struct2025.11 | 77.6 | 77.8 | 77.2 | |
| GPT-4.1-miniPipeline=VLM + Tesseract2025.11 | 77.5 | 81.6 | 79 | |
| Qwen-7BPipeline=Only VLM2025.11 | 77.4 | 72.7 | 74.4 | |
| GPT-4oPipeline=VLM + Tesseract2025.11 | 77.3 | 77.8 | 77.2 | |
| Mistral-small-3.1Pipeline=VLM + RapidOCR2025.11 | 77.3 | 78.8 | 77.6 | |
| GPT-4.1-miniPipeline=VLM + Pix2Struct2025.11 | 77.2 | 81.6 | 78.8 | |
| GPT-4oPipeline=VLM + RapidOCR2025.11 | 76.9 | 78 | 77.1 | |
| Qwen-7BPipeline=VLM + Pix2Struct2025.11 | 76.9 | 74.4 | 74.9 | |
| GPT-4oPipeline=VLM + Pix2Struct2025.11 | 76.7 | 78.1 | 77 | |
| GPT-4.1-miniPipeline=VLM + RapidOCR2025.11 | 76.6 | 81.4 | 78.4 | |
| Gemma-12BPipeline=VLM + Pix2Struct2025.11 | 74.8 | 56 | 62.8 | |
| GPT-4o-miniPipeline=VLM + Pix2Struct2025.11 | 74.8 | 63.3 | 67.8 | |
| GPT-4o-miniPipeline=VLM + RapidOCR2025.11 | 74.8 | 63.3 | 67.8 | |
| GPT-4o-miniPipeline=VLM + Tesseract2025.11 | 73.9 | 64.2 | 68.1 | |
| Qwen-7BPipeline=VLM + Tesseract2025.11 | 73.6 | 73.8 | 72.9 | |
| Qwen-7BPipeline=VLM + RapidOCR2025.11 | 73.5 | 73.8 | 72.9 | |
| Gemma-12BPipeline=VLM + RapidOCR2025.11 | 72.5 | 59.2 | 64 | |
| Gemma-12BPipeline=VLM + Tesseract2025.11 | 72.5 | 59.2 | 64 | |
| Pixtral-largePipeline=VLM + RapidOCR2025.11 | 71.6 | 51.5 | 58.9 | |
| Pixtral-largePipeline=VLM + Tesseract2025.11 | 71.5 | 51.6 | 58.7 | |
| GPT-4.1-nanoPipeline=Only VLM2025.11 | 71.2 | 53.5 | 59.8 | |
| Pixtral-largePipeline=Only VLM2025.11 | 71.2 | 51.7 | 58.9 | |
| Pixtral-largePipeline=VLM + Pix2Struct2025.11 | 70.6 | 51.7 | 58.3 | |
| GPT-4.1-nanoPipeline=VLM + RapidOCR2025.11 | 68.1 | 52.2 | 57.9 | |
| GPT-4.1-nanoPipeline=VLM + Pix2Struct2025.11 | 66.1 | 52.9 | 57.6 | |
| GPT-4.1-nanoPipeline=VLM + Tesseract2025.11 | 64.5 | 50.4 | 55.3 | |
| Gemma-4BPipeline=Only VLM2025.11 | 64.2 | 58.6 | 60.2 | |
| Gemma-4BPipeline=VLM + Pix2Struct2025.11 | 63.5 | 58.6 | 59.9 | |
| Gemma-4BPipeline=VLM + RapidOCR2025.11 | 61.7 | 57.6 | 58.5 | |
| Gemma-4BPipeline=VLM + Tesseract2025.11 | 61.7 | 57.6 | 58.5 | |
| Qwen-3BPipeline=VLM + RapidOCR2025.11 | 56.8 | 40.8 | 45.8 | |
| Qwen-3BPipeline=VLM + Pix2Struct2025.11 | 55 | 46.9 | 48.1 | |
| Qwen-3BPipeline=VLM + Tesseract2025.11 | 54.7 | 40.6 | 45.1 | |
| Qwen-3BPipeline=Only VLM2025.11 | 53.8 | 48.6 | 48.6 | |
| BPMN RedrawerPipeline=Non-VLM Baseline2025.11 | 53.1 | 40.7 | 45.8 | |
| LLaMA-3.2-11BPipeline=VLM + RapidOCR2025.11 | 39.4 | 32.7 | 34.3 | |
| LLaMA-3.2-11BPipeline=VLM + Pix2Struct2025.11 | 37.5 | 30.2 | 32.2 | |
| LLaMA-3.2-11BPipeline=Only VLM2025.11 | 37.3 | 33.4 | 34.4 | |
| Pixtral-12BPipeline=VLM + Tesseract2025.11 | 35.4 | 31.2 | 32.3 | |
| Pixtral-12BPipeline=Only VLM2025.11 | 35.2 | 28.4 | 30.1 | |
| LLaMA-3.2-11BPipeline=VLM + Tesseract2025.11 | 34.4 | 30.9 | 31.8 | |
| Pixtral-12BPipeline=VLM + Pix2Struct2025.11 | 34.2 | 31.2 | 31.6 | |
| Pixtral-12BPipeline=VLM + RapidOCR2025.11 | 32.1 | 29.2 | 29.7 | |
| Aya-vision-8BPipeline=VLM + RapidOCR2025.11 | 21.3 | 13.3 | 15.4 | |
| Aya-vision-8BPipeline=Only VLM2025.11 | 20.1 | 14 | 15.1 | |
| Aya-vision-8BPipeline=VLM + Tesseract2025.11 | 19 | 12.9 | 14.2 | |
| Aya-vision-8BPipeline=VLM + Pix2Struct2025.11 | 17.5 | 12.8 | 13.8 |