A Short Note on the Kinetics-700 Human Action Dataset

2019/07/15 by João Carreira, Joao Carreira, Eric Noland +6 · 53 citations
Computer Science · #Anomaly Detection Techniques and Applications #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Human Pose and Action Recognition #cs.CV

paper · pdf · doi:10.48550/arxiv.1907.06987

added note about dangers of training on k700 and evaluating on k400/k600. arXiv admin note: text overlap with arXiv:1808.01340

openalex publication_date 2019/07/15 · arxiv created 2022/10/17 · arxiv updated 2022/10/19 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

We describe an extension of the DeepMind Kinetics human action dataset from 600 classes to 700 classes, where for each class there are at least 600 video clips from different YouTube videos. This paper details the changes introduced for this new release of the dataset, and includes a comprehensive set of statistics as well as baseline results using the I3D neural network architecture.

Citations

• A Short Note about Kinetics-600

Cited by

• Open Ad-hoc Categorization with Contextualized Feature Learning
• Recurrent Video Masked Autoencoders
• Robust Motion Generation using Part-level Reliable Data from Videos
• Generative Action Tell-Tales: Assessing Human Motion in Synthesized Videos
• InternVideo-Next: Towards General Video Foundation Models without Video-Text Supervision
• ViMix-14M: A Curated Multi-Source Video-Text Dataset with Long-Form, High-Quality Captions and Crawl-Free Access
• EventBench: Towards Comprehensive Benchmarking of Event-based MLLMs
• A Step Toward World Models: A Survey on Robotic Manipulation
• LongCat-Flash-Omni Technical Report
• Towards Universal Video Retrieval: Generalizing Video Embedding via Synthesized Multimodal Pyramid Curriculum
• CG-World: A Large-Scale World-State Dataset and Protocol for World Models
• OpenMAP-BrainAge: generalizable and interpretable brain age predictor from MRI
• A Comprehensive Survey on World Models for Embodied AI
• StretchySnake: Flexible SSM Training Unlocks Action Recognition Across Spatio-Temporal Scales
• Image-to-Video Transfer Learning based on Image-Language Foundation Models: A Comprehensive Survey
• DynamicEval: Rethinking Evaluation for Dynamic Text-to-Video Synthesis
• Vision-Language-Action Models for Robotics: A Review Towards Real-World Applications
• UniVid: The Open-Source Unified Video Model
• CoWTracker: Tracking by Warping instead of Correlation
• InterPose: Learning to Generate Human-Object Interactions from Large-Scale Web Videos
• No More Sibling Rivalry: Debiasing Human-Object Interaction Detection
• Category-level Text-to-Image Retrieval Improved: Bridging the Domain Gap with Diffusion Models and Vision Encoders
• Representation Learning with Adaptive Superpixel Coding
• Survey of Vision-Language-Action Models for Embodied Manipulation
• MOVER: Multimodal Optimal Transport with Volume-based Embedding Regularization
• Text as Any-Modality for Zero-Shot Classification by Consistent Prompt Tuning
• iSafetyBench: A video-language benchmark for safety in industrial environment
• Enhancing Scene Transition Awareness in Video Generation via Post-Training
• CoTasks: Chain-of-Thought based Video Instruction Tuning Tasks
• SPACT18: Spiking Human Action Recognition Benchmark Dataset with Complementary RGB and Thermal Modalities
• DVFL-Net: A Lightweight Distilled Video Focal Modulation Network for Spatio-Temporal Action Recognition
• V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and Planning
• Language-Guided Contrastive Audio-Visual Masked Autoencoder with Automatically Generated Audio-Visual-Text Triplets from Videos
• Simplifying Traffic Anomaly Detection with Video Foundation Models
• VLM2Vec-V2: Advancing Multimodal Embedding for Videos, Images, and Visual Documents
• Grounding Intelligence in Movement
• Improving Progressive Generation with Decomposable Flow Matching
• OpenMAP-BrainAge: Generalizable and Interpretable Brain Age Predictor
• With Limited Data for Multimodal Alignment, Let the STRUCTURE Guide You
• EchoShot: Multi-Shot Portrait Video Generation
• Bootstrapping World Models from Dynamics Models in Multimodal Foundation Models
• ExAct: A Video-Language Benchmark for Expert Action Analysis
• Person-In-Situ: Scene-Consistent Human Image Insertion with Occlusion-Aware Pose Control
• Large-scale Self-supervised Video Foundation Model for Intelligent Surgery
• MotionSight: Boosting Fine-Grained Motion Understanding in Multimodal LLMs
• V-JEPA 2.1: Unlocking Dense Features in Video Self-Supervised Learning
• ActionArt: Advancing Multimodal Large Models for Fine-Grained Human-Centric Video Understanding
• OmniVDiff: Omni Controllable Video Diffusion for Generation and Understanding
• Mavors: Multi-granularity Video Representation for Multimodal Large Language Model
• SEVERE++: Evaluating Benchmark Sensitivity in Generalization of Video Representation Learning
• SVLTA: Benchmarking Vision-Language Temporal Alignment via Synthetic Video Situation
• A Large-Scale Analysis on Contextual Self-Supervised Video Representation Learning
• Scaling Open-Vocabulary Action Detection

Related

• A Short Note on the Kinetics-700-2020 Human Action Dataset
• A Short Note about Kinetics-600
• The Kinetics Human Action Video Dataset
• Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset
• The AVA-Kinetics Localized Human Actions Video Dataset