Claude-3 Sonnet

Reported on 3 benchmarks across 1 task · 1 paper · 2 SOTA

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Methodology3 results

Optical Character Recognition (OCR)onVideoDB's OCR Benchmark Public Collection
Character Error Rate (CER)· 2025-02-10
0.3229
best: 0.2378 (GPT-4o)
SOTA
Benchmarking Vision-Language Models on Optical Character Recognition in Dynamic Video Environments arXiv:2502.06445
Optical Character Recognition (OCR)onVideoDB's OCR Benchmark Public Collection
Word Error Rate (WER)· 2025-02-10
0.4663
best: 0.2385 (Gemini-1.5 Pro)
SOTA
Benchmarking Vision-Language Models on Optical Character Recognition in Dynamic Video Environments arXiv:2502.06445
Optical Character Recognition (OCR)onVideoDB's OCR Benchmark Public Collection
Average Accuracy· 2025-02-10
67.71
best: 76.22 (GPT-4o)
Benchmarking Vision-Language Models on Optical Character Recognition in Dynamic Video Environments arXiv:2502.06445