🐉 PaddleOCR Benchmark (PP-OCRv5 — PaddlePaddle Runtime)
Benchmark PaddleOCR with the full PaddlePaddle inference runtime on public OCR datasets.
| Property | Value |
|---|---|
| Engine | PaddleOCR 3.5+ |
| Model version | PP-OCRv5 (latest) |
| Runtime | PaddlePaddle (native) |
| AWS Lambda zip (250 MB)? | ❌ Exceeds limit |
| AWS Lambda container (10 GB)? | ✅ Fits |
📏 Deployment sizes are measured from actual installed files on disk — not pip metadata.
💡 Compare with the RapidOCR benchmark Space to see how ONNX Runtime reduces deployment size while preserving accuracy.
Run PaddleOCR on a benchmark dataset and measure accuracy, speed & deployment footprint.
Select Benchmark Dataset
5 50
Metrics