AnandMayank/QueST-PartNetMobility-SAPIEN
QueST: PartNet-Mobility SAPIEN Simulation Dataset This dataset accompanies the paper: QueST: Persistent Queries as Semantic Monitors for Drift Suppression in Long-Horizon TrackingMayank Anand, Mohammad Saqlain, Kyan Mahajan, Priya Shukla, G.C Nandi, Andrew MelnikCAO Workshop at ICLR 2026 What Is This Dataset? Synchronized RGB-D simulation sequences rendered in SAPIEN from PartNet-Mobility articulated objects, designed to stress-test long-horizon point… See the full description on the dataset page: https://huggingface.co/datasets/AnandMayank/QueST-PartNetMobility-SAPIEN.
🎯 QueST: Bộ dữ liệu mô phỏng RGB-D theo dõi điểm dài hạn cho đối tượng articulated
Bộ dữ liệu RGB-D mô phỏng SAPIEN từ PartNet-Mobility, thiết kế để test truy xuất điểm dài hạn dưới stress từ artulation, occlusion và viewpoint change.
QueST-PartNetMobility-SAPIEN là bộ dữ liệu đồng bộ RGB-D được render bởi SAPIEN từ các đối tượng articulated của PartNet-Mobility, hỗ trợ khung QueST — thay thế truy xuất Markov truyền thống bằng truy vấn ngữ nghĩa bền vững, giảm 67.7% lỗi APE trên chuỗi articulated dài hạn so với TAP-Net.
✨ Điểm nổi bật
- RGB-D đồng bộ 3 góc nhìn: Hỗ trợ đánh giá đa góc và huấn luyện mô hình thời gian thực.
- 4 cấp độ độ phức tạp: Từ manipulation_1 (1 khớp) đến manipulation_4 (4 khớp, 240 khung).
- Chú thích affordance pixel-level: Nhãn vùng tương tác ở mỗi khung hình.
- Trajectory GT 3D: Đường mặt bản đồ điểm chính xác cho đánh giá.
- Metadata chi tiết: Bao gồm joint config, camera intrinsics, trajectory GT.
- Hỗ trợ video + frames PNG: Linh hoạt cho cả huấn luyện video và đánh giá khung.
- 50,551 ảnh, 5.75GB: Quy mô lớn cho bài toán tracking dài hạn.
💡 Vì sao đáng chú ý
Khác với bộ dữ liệu tracking ngắn hạn, QueST tập trung vào stress test dài hạn với artulation liên tục — nơi các tracker Markov thường bị drift. QueST framework dùng truy vấn ngữ nghĩa bền vững thay vì truyền thống frame-to-frame, giảm đáng kể lỗi trên chuỗi dài. Phù hợp cho nghiên cứu long-horizon point tracking, articulated object manipulation và embodied AI.
#QueST #PartNetMobility #SAPIEN #PointTracking #LongHorizon #ArticulatedObjects #TiniX
Similar Projects
Discover other similar projects and AI models.
Stera-10M Visualizer: https://platform.fpvlabs.ai/dataset/stera-10m/viz Dataset Summary Stera-10M is an open egocentric multimodal dataset for embodied AI, robotics, world models, and spatial intelligence, captured end-to-end on commodity iPhone Pro hardware through the open Stera platform. It contains 200 hours of synchronized first-person recordings across 500+ sessions from 20 contributors in 20+ unique environments, with 10 million RGB frames, LiDAR depth… See the full description on the dataset page: https://huggingface.co/datasets/fpvlabs/stera-10m.
Dataset Card for Street Videos Dataset This dataset contains a collection of street videos designed to support tasks such as video classification, object detection, and urban scene analysis. Each video captures various street environments, including urban settings, pedestrian areas, and traffic scenarios, to provide diverse training data for computer vision models. Dataset Details Dataset Description This dataset contains videos of street scenes… See the full description on the dataset page: https://huggingface.co/datasets/HumynLabs/Street-videos.
EgoPro The 10,000-hour head-and-wrist line of EgoSuite-Open100K. Data Bucket · Collection · EgoDemo · EgoStandard · Project page Explore EgoSuite-Open100K ↗ Data location: EgoPro is distributed through the LightwheelAI/EgoPro Bucket. This Git repository is the dataset card and access point; download the data from the Bucket. Overview EgoPro pairs synchronized head- and wrist-view video with 3D hand pose. Its body subset adds full-body pose. LeRobot and… See the full description on the dataset page: https://huggingface.co/datasets/LightwheelAI/EgoPro.
⚠️ Important: If you have already submitted an access request but have not completed the required DocuSign agreement, your request will remain pending. Please complete signing and we will grant access once verified. Interactive Intelligence from Human Xperience Xperience-10M Dataset Summary Xperience-10M is a large-scale egocentric multimodal dataset of human experience for embodied AI, robotics, world models, and spatial… See the full description on the dataset page: https://huggingface.co/datasets/ropedia-ai/xperience-10m.
all-MiniLM-L6-v2 This is a sentence-transformers(https://www.SBERT.net) model: It maps sentences & paragraphs to a 384 dimensional dense vector space and can be used for tasks like clustering or semantic search. Usage (Sentence-Transformers) Using t...
Dataset Card for Car Images Dataset Dataset Description This dataset contains a collection of car images designed to support tasks such as image classification, car detection, and autonomous driving research. The images feature cars in various settings, including streets and other environments, captured to provide diverse training data for computer vision models. The dataset aims to enable the development and benchmarking of models that can: Identify… See the full description on the dataset page: https://huggingface.co/datasets/HumynLabs/car-images.
EgoDemo A 50-hour sample from EgoSuite-Open100K, covering every annotated subset plus two raw-video variants. Collection · EgoStandard · EgoPro · Project page Explore EgoSuite-Open100K ↗ EgoSuite-Open100K Overview Collection: EgoSuite-Open100K SKU Sub-SKU Format Planned Duration EgoStandard EgoStand Hand Pose 80,000 h… See the full description on the dataset page: https://huggingface.co/datasets/LightwheelAI/EgoDemo.
EgoStandard The 90,000-hour head-view line of EgoSuite-Open100K. Data Bucket · Collection · EgoDemo · EgoPro · Project page Explore EgoSuite-Open100K ↗ Data location: EgoStandard is distributed through the LightwheelAI/EgoStandard Bucket. This Git repository is the dataset card and access point; download the data from the Bucket. Overview EgoStandard pairs head-view egocentric video with synchronized 3D hand pose. Its body subset adds full-body pose.… See the full description on the dataset page: https://huggingface.co/datasets/LightwheelAI/EgoStandard.
Eidon Tracker POV 1,274 hours of egocentric video paired with 7-point IMU arm tracking, recorded during ordinary household work. Contributors wore a head-mounted camera and a seven-sensor IMU harness while doing real chores in their own homes: laundry, cleaning, dishes, cooking. Each recording pairs first-person video with 24 Hz orientation data for both hands, both forearms, both upper arms, and the chest. This is a complete, final release. Eidon AI (Solidic Labs Inc) has wound… See the full description on the dataset page: https://huggingface.co/datasets/eidon-ai/tracker-pov.