20 datasets found

Object permanence is the concept that objects in a physical world continue to exist despite the observers inability to sense them. This can result in temporally unstable...

Person re-identification (re-ID) under various occlusions has been a long-standing challenge as person images with different types of occlusions often suffer from misalign-ment...

The dataset used in the paper is a video prediction dataset with occlusions, which is used to evaluate the proposed Fast Fourier Inception Networks (FFINet) for occluded video...

The dataset used in the paper is a video prediction dataset with occlusions, which is used to evaluate the proposed Fast Fourier Inception Networks (FFINet) for occluded video...

The Stanford Cars dataset is a dataset of images of cars, with 196 categories and approximately 16,000 images. The authors created a synthetic dataset by adding occlusions of...

A new dataset for multi-person 3D pose estimation with occlusion handling in moving camera environments.

The DriveU Traffic Light Dataset is a dataset of annotated traffic lights with various features, including color, directionality, and occlusion level.

LINEMOD dataset is a widely used instance-level 6D object pose estimation dataset which consists of 13 different objects with significant shape variation.

The proposed method employs a divide-and-conquer strategy for the ReID task. Concretely, two self-supervision operations are used to generate new images with the characteristics...

Face recognition dataset with occlusions

The dataset used in the FairFace Challenge at ECCV 2020 is a reannotated version of IJB-C [37] database enriched by newly collected 12,549 public domain images.

The IJB-A dataset is a large-scale unconstrained face recognition dataset, containing 5,397 images and 2,042 videos from 500 subjects, captured from in-the-wild environment to...

The CMU-Panoptic dataset is a multi-person indoor dataset collected with multi-view cameras. The dataset is used for occluded human mesh recovery.

The 3DPW-Crowd dataset is a person crowded subset of 3DPW, containing 1923 persons. The dataset is used for occluded human mesh recovery.

Estimating 3D human pose and shape from monocular images is a crucial task for various applications such as performance retargeting, virtual avatars, and human action recognition.

The 3DPW-PC dataset is a person-person occlusion subset of 3DPW, containing 2218 individuals. The dataset is used for occluded human mesh recovery.

The 3DPW-OC dataset is a person-object occlusion subset of 3DPW, containing 20243 persons. The dataset is used for occluded human mesh recovery.

The dataset used for 6D object pose estimation, consisting of images of 16 objects with varying levels of occlusion.

The OCHuman dataset is a more challenging dataset, where each human instance is heavily occluded by one or several others and the postures of the human bodies are more complex.

The MNIST dataset (Lecun et al., 1998), which consists of 60,000 gray-scale images of handwritten digits. Each image has an accompanying label in {0, 1,..., 9}, and is stored as...