69 datasets found

The PersonLab dataset is a benchmark for person pose estimation and instance segmentation. It contains 20,000 images with 80,000 annotated person instances.

Uni3DL is a unified model for 3D and language understanding. It operates directly on point clouds and supports diverse 3D vision-language tasks, including semantic segmentation,...

The MOTS20 dataset is used for multi-object tracking and segmentation.

The OpenImages dataset V6 contains images with annotations for object detection and instance segmentation tasks.

The SUNRGB-D dataset features 5,285 training and 5,050 test samples from multiple RGB-D cameras. The dataset provides annotations for the first 37 NYUv2 semantic classes and for...

Amodal intra-class instance segmentation: Synthetic datasets and benchmark.

Amodal intra-class instance segmentation: Synthetic datasets and benchmark.

Amodal intra-class instance segmentation: Synthetic datasets and benchmark.

The dataset used for one-shot semantic segmentation, including COCO-20i, FSS-1000, and LVIS-92i.

Po-LarMask: Single shot instance segmentation with polar representation.

Solo: Segmenting objects by locations.

Microsoft COCO: common objects in images.

Instance segmentation is one of the fundamental vision tasks. Recently, fully convolutional instance segmentation methods have drawn much attention as they are often simpler and...

The dataset used in the paper is the COCO object detection and instance segmentation dataset, and the ADE20K semantic segmentation dataset.

The COCO 2014 validation set is used for evaluating the performance of the proposed method.

Fashionformer is a unified baseline for human fashion segmentation and recognition, achieving state-of-the-art performance on Fashionpedia.

Fashionpedia is a widely recognized dataset for fashion parsing, consisting of images of fashion items with annotations.

FashionFail is a dataset for precise fashion object detection and segmentation, serving as a robustness benchmark for fashion parsing models trained on “in-the-wild” images such...

Yolact dataset contains 100,000 images with 500,000 annotated instances.

The dataset used in the paper is ScanNetV2, a real-world dataset for indoor scenes. It contains 1205 training scenes and 312 testing scenes, with instance-level object bounding...