Optical Character Recognition (OCR) on VideoDB's OCR Benchmark Public Collection

Metric: Character Error Rate (CER) (lower is better)

LeaderboardDataset

Loading chart...

Results

Submit a result

Sort:

#	Model↕	Character Error Rate (CER)▲	Augmentations	Paper	Date↕	Code
1	GPT-4o	0.2378	No	Benchmarking Vision-Language Models on Optical C...	2025-02-10	Code
2	Gemini-1.5 Pro	0.2387	No	Benchmarking Vision-Language Models on Optical C...	2025-02-10	Code
3	Claude-3 Sonnet	0.3229	No	Benchmarking Vision-Language Models on Optical C...	2025-02-10	Code
4	EasyOCR	0.507	No	Benchmarking Vision-Language Models on Optical C...	2025-02-10	Code
5	RapidOCR	0.762	No	Benchmarking Vision-Language Models on Optical C...	2025-02-10	Code

#1GPT-4oSOTA
0.2378
Character Error Rate (CER)· 2025-02-10
Benchmarking Vision-Language Models on Optical Character Recognition in Dynamic Video Environments Code
#2Gemini-1.5 ProSOTA
0.2387
Character Error Rate (CER)· 2025-02-10
Benchmarking Vision-Language Models on Optical Character Recognition in Dynamic Video Environments Code
#3Claude-3 SonnetSOTA
0.3229
Character Error Rate (CER)· 2025-02-10
Benchmarking Vision-Language Models on Optical Character Recognition in Dynamic Video Environments Code
#4EasyOCRSOTA
0.507
Character Error Rate (CER)· 2025-02-10
Benchmarking Vision-Language Models on Optical Character Recognition in Dynamic Video Environments Code
#5RapidOCRSOTA
0.762
Character Error Rate (CER)· 2025-02-10
Benchmarking Vision-Language Models on Optical Character Recognition in Dynamic Video Environments Code