120 datasets found

A new challenging dataset for real-world object picking tasks, consisting of 1,180 images with bounding boxes and text instructions annotated.

The UT-Zap50K dataset is a dataset of images of shoes.

A wearable assistive device for visually impaired people to navigate safely and quickly in unfamiliar environments, as well as perceive the complicated environment to...

A dataset of video sequences from [6].

We use the YouTube-Objects dataset for video object segmentation.

The MonuMAI dataset is a part-based object recognition dataset for classifying architectural styles of monument facades.

A dataset of 10 simple object classes (pineapple, soccer, basketball, etc.) with white backgrounds.

The McGill dataset is a 3D point cloud dataset containing 3D models of various objects.

Exploit bounding box annotations for multi-label object recognition.

A dataset of 15,339 images with 64 attributes, used for Attribute Pascal and Yahoo (APY) dataset.

The Taskonomy dataset contains 16 types of annotation, including object classification, scene classification, depth estimation, surface normals, curvature estimation, edge...

CLOW is a dataset collected via a mixed-reality interactive approach for continual learning about objects in the wild.

PascalVOC 2007 dataset consists of 5011 training image-text pairs and 4952 test image-text pairs corresponding to 20 different object categories.

The multi-view multi-class dataset contains 23 minutes and 57 seconds of synchronized frames taken at 25fps from 6 different calibrated DV cameras.

A dataset for training a Neural Sparse Voxel Fields (NSVF) model, containing 3D images of objects.

Scanned objects dataset

Abo: Dataset and benchmarks for real-world 3D object understanding

The Playroom is a Unity-based environment for object-centric tasks [69, 46], originally released as pre-packaged Docker containers with an Apache 2.0 license.

Objects Room was a MuJoCo-based dataset originally released under the Apache 2.0 license and used for prior work such as GQN and MONet.

The iCubWorld dataset is a benchmark for object recognition in robotics. It consists of 150 objects evenly divided into 15 categories with their appearance changing due to...