Document Rotation Classification on ORB-En
98AccuracyPhi-3.5-Vision rotation classification module
Evaluation Results
| Method | Links | |
|---|---|---|
| Phi-3.5-Vision rotation classification moduleEvaluation protocol=Proposed2025.11 | 98 | |
| Phi-3.5-Vision rotation classification module2025.11 | 96.81 | |
| TrOCREvaluation protocol=fine-tuned2025.11 | 78.18 | |
| LLaMA-4 (Maverick)Parameters=17B2025.11 | 65.49 | |
| docTR2025.11 | 63.28 | |
| GPT-4o2025.11 | 59.58 | |
| Gemini-2.5 Flash2025.11 | 39.55 | |
| Gemini-2.5 Pro2025.11 | 34.11 | |
| GPT-4oEvaluation protocol=0-shot2025.11 | 34 | |
| LLaMA-4 (Maverick) 17BEvaluation protocol=0-shot2025.11 | 32 | |
| Gemma-3Parameters=27B2025.11 | 30.57 | |
| TrOCR2025.11 | 26.22 | |
| Nanonets-OCR-s2025.11 | 24.83 | |
| Gemini-2.5 ProEvaluation protocol=0-shot2025.11 | 24 | |
| Gemini-2.5 FlashEvaluation protocol=0-shot2025.11 | 22 | |
| Gemma-3 27BEvaluation protocol=0-shot2025.11 | 16 | |
| docTREvaluation protocol=fine-tuned2025.11 | 10 | |
| Nanonets-OCR-sEvaluation protocol=0-shot2025.11 | 8 |