Egocentric Benchmarks

Created in July 04, 2025

2025   ·   dataset   ·   dataset

Action Recognition & Activity Understanding

  • EPIC-KITCHENS: Unscripted cooking activities with object interactions, also involves EPIC Fields and EPIC-SOUNDS
  • Ego4D: Large-scale daily activities (1,400+ hrs, 2.2M annotations), it also has EgoTracks as a part.
  • EGTEA Gaze: Kitchen activities with gaze tracking
  • EGTEA Gaze+: Another daatset with Gaze Tracking. There is GTEA Gaze, Gaze+, sub, EGTEA Gaze+
  • Charades-Ego: Paired third-person and egocentric daily activities
  • ASCC Activities of Daily Living: Recorded with a chest-mounted camera, showing activities of daily living such as cooking, cleaning, etc.
  • EgoPet: From pet based egocentric view
  • AEA: Everyday Activity dataset
  • DataEgo: egocentric dataset composed of visual, gyroscope and accelerometer information
  • EgoPER: Egocentric Procedural Task Videos
  • WEAR: Wearable and Egocentric Activity Recognition
  • IndustReal: Action recognition and assembly state detection
  • EgoProceL: Dataset for procedural learning

Hand Interaction

  • EgoHands: Hand detection and segmentation
  • EgoDexter
  • H2O: Human-human-object interactions
  • EgoGesture: multi-modal large scale dataset for egocentric hand gesture recognition
  • EgoChoir: estimate 3D human contact and obejct affordance from egocentric videos
  • HOT3D: 3D hand and object tracking
  • ARCTIC: Dexterous Bimanual Hand-Object Manipulation
  • HOI4D: Category-Level Human-Object Interaction

Long-Term Video Understanding

  • DailyLife: Long-term activity recognition (12 hours/day)
  • EgoRoutine: Routine understanding over extended periods

Social Interaction

  • EgoCom: Multi-person conversations
  • EgoReID: Person Reidentification
  • EgoHumans: Egocentric human 3D pose estimation and tracking
  • VizWiz-Ego: Assistive vision for visually impaired users
  • Ego-Centric: Object interactions in daily environments

Gaze and Attention Prediction

  • EHTask: Gaze behavior during manipulation tasks

Motion and Pose Estimation

  • EgoCap: 3D body pose from head-mounted cameras
  • xr-EgoPose: Egocentric 3D human pose estimation
  • Nymeria: Human Pose Estimation
  • Ego-Exo4D: Meticulously synchronized natural language datasets paired with videos
  • EgoExoLearn: Asynchronous Ego- and Exo-centric View of Procedural Activities in Real World
  • Assembly101: Multi-View Video Dataset for Understanding Procedural Activities
  • Mo2Cap2: Egocentric estimation of 3D human body pose in a wide range of unconstrained everyday activities
  • EgoWholeMocap: Egocentric whole-body motion capture using a single fisheye camera, which simultaneously estimates human body and hand motion
  • SceneEgo: Scene-aware Egocentric 3D Human Pose Estimation
  • EE3D: Monocular egocentric 3D human motion capture
  • EgoBody3M: Accurate tracking of a user’s body pose while wearing a virtual reality (VR), augmented reality (AR) or mixed reality (MR)
  • AssemblyHands: 3D Hand Pose Estimation

3D Human Motion Capture

  • UnrealEgo: Egocentric 3D Human Motion Capture
  • UnrealEgo2: Stereo egocentric 3D human pose estimation
  • FPHA: RGB-D Videos and 3D Hand Pose Annotations

Object Detection

  • TACO: Image dataset of waste in the wild
  • HInt: Reconstructing Hands in 3D with Transformers
  • ADL: detecting activities of daily living (ADL) in first-person camera views

Image Based Localization

  • EgoCART: Indoor Image-Based Localization
* indicates required

Intuit Mailchimp