VideoDB leads a 100-hour action retrieval benchmark vs. AWS Nova and Twelve Labsvideodb.io·2 pts·ashu_trv·0
Show HN: Do Thought Streams Matter? A Benchmark of VLM Reasoning in Gemini 2.5arxiv.org·3 pts·ashu_trv·0
Benchmarking vision-language models on OCR in dynamic video environmentsarxiv.org·142 pts·ashu_trv·58