Kinetics-400

Kinetics-400 is a benchmark dataset that contains a diverse set of video clips for evaluating fine-grained video understanding and reasoning tasks.

Papers using Kinetics-400 (6)

Video-STAR: Reinforcing Open-Vocabulary Action Recognition with Tools2025ReasonAct: Progressive Training for Fine-Grained Video Reasoning in Small Models2025Q-CLIP: Unleashing the Power of Vision-Language Models for Video Quality Assessment through Unified Cross-Modal Adaptation2025Bidirectional Cross-Modal Knowledge Exploration for Video Recognition with Pre-trained Vision-Language Models2023 · 6 citesVicTR: Video-conditioned Text Representations for Activity Recognition2023Mug-STAN: Adapting Image-Language Pretrained Models for General Video Understanding2023