246 datasets found

Polyp segmentation using U-Net-ResNet50

The Class Incremental Semantic Segmentation (CISS) extends the traditional seg-mentation task by incrementally learning newly added classes.

A reinterpretation of the U-Net architecture as a modularized structure was proposed as a solution to scale up the segmentation of such images.

Pyramid scene parsing network for semantic segmentation.

The MS COCO dataset is a large benchmark for image captioning, containing 328K images with 5 caption descriptions each.

The COCO+LVIS dataset contains millions of high-quality labels for natural images.

G-Ref is a dataset for referring image segmentation, comprising 104K referring language expressions for around 55K objects in about 27K images.

A public PSV dataset for parking slot and lane markings segmentation

The MNIST(multi) dataset is used to test the ability of TARGET-VAE to identify multiple objects in images.

The proposed pipeline, VertXNet, is trained and tested on either anonymized X-ray images (MEASURE 1 and PREVENT) from completed and anonymized secukinumab clinical trials or a...

The COD10K dataset is currently the largest challenging dataset for COD, containing 10K images with dense annotations.

The dataset used for camouflaged object detection, shadow detection, and polyp segmentation tasks.

The dataset used for training and testing the OMG-Seg model, which includes COCO panoptic, COCO-SAM, VIPSeg, Youtube-VIS-2019, and Youtube-VIS-2021 datasets.

The dataset used in this paper for multi-sensor next-best-view planning as matroid-constrained submodular maximization.

Brain tumor segmentation dataset used for training and evaluation of the proposed Squeeze Excitation Embedded Attention UNet (SEEA-UNet) model.

The RIM-ONE-r3 dataset contains 60 test images.

The Refuge challenging data set contains 400 train, 400 validation, and 400 test images.

Convex shapes are common priors for optic disc and cup segmentation in eye fundus images. This method is proposed to guarantee that outputs are convex shapes.

DUT is a new defocus blur detection dataset which consists of 500 images as the test set and 600 images as the train set.

CUHK is a classical blur detection dataset, among which 296 images are partially motion-blurred, and 704 images are defocus-blurred.