Computer Vision
This dataset accompanies the paper “Evaluating Cross-Device and Cross-Subject Consistency in Visual Fixation Prediction”. We collected eye gaze data using a 30Hz eye tracker embedded in the Aria Glasses (Meta Platforms, Inc., Menlo Park, CA, USA) on 300 images from the MIT1003 dataset, with each image viewed for 3 seconds by 9 subjects (age range 23-39 years), resulting in a total of 243,000 eye fixations. Besides, we also release the average saliency maps from the subjects' visual fixations.
- Categories:
111 Views
The BEVTrack benchmark consists of two complementary datasets: CVMHAT-BEVT and BEVT-S, designed for multi-view human tracking with bird's-eye view (BEV) capabilities. CVMHAT-BEVT is adapted from the public CVMHAT dataset, featuring 21 synchronous multi-view videos across five scenes, with 7-12 individuals per scene and video durations ranging from 200 to 1,500 frames. Each scene contains 2-4 side views and includes synchronized BEV footage captured by drones, along with annotated bounding boxes and unified ID numbers for all subjects across views.
- Categories:
12 Views
The turbid dataset is specifically designed to support research in riverine underwater image enhancement (UIE), addressing the challenges posed by muddy underwater conditions. It comprises three distinct transformed datasets: Turbid-UIEB, Turbid-LSUI, and Turbid-ImageNet, each created through a diffusion model technique. Turbid-UIEB features 890 paired images. Turbid-LSUI expands upon this with 4,278 paired images, offering a broader range of examples for training and testing.
- Categories:
149 Views
CIFAR-10 and CIFAR-100 datasets comprise images of 10 and 100 categories, respectively, with a fixed size of 32x32 pixels in color.
Tiny-ImageNet dataset consists of 200 categories with approximately 120,000 samples, where each class contains 500 training images, 50 validation images, and 50 test images, with each image sized at 64 x 64.
AG News dataset consists of article titles and descriptions, comprising 4 categories with 127,600 samples.
- Categories:
13 Views
Dataset for PerfCam: Digital Twinning for Production Lines Using 3D Gaussian Splatting and Vision Models
KTH Royal Institute of Technology, SCI; AstraZeneca, Sweden Operations
- Categories:
155 Views
AH-Rain was collected within Anhui Province, China, covering a spatial range from 27°N to 36°N latitude and from 113°E to 121°E longitude. It has a temporal resolution of 6 minutes and a spatial resolution of 0.01° × 0.01°, with each radar echo image having a resolution of 800 × 900 pixels. S-band radar was used to collect data at nine elevation angles per volume scan. The composite reflectivity, which represents the maximum value among the nine elevation angles at the same azimuth and distance, was obtained. The data was collected over the years 2021, 2022, and 2023.
- Categories:
9 Views
A new small aerial flame dataset, called the Aerial Fire and Smoke Essential (AFSE) dataset, is created which is comprised of screenshots from different YouTube wildfire videos as well as images from FLAME2. Two object categories are included in this dataset: smoke and fire. The collection of images is made to mostly contain pictures utilizing aerial viewpoints. It contains a total of 282 images with no augmentations and has a combination of images with only smoke, fire and smoke, and no fire nor smoke.
- Categories:
350 Views
This dataset comprises 33,800 images of underwater signals captured in aquatic environments. Each signal is presented against three types of backgrounds: pool, marine, and plain white. Additionally, the dataset includes three water tones: clear, blue, and green. A total of 12 different signals are included, each available in all six possible background-tone combinations.
- Categories:
329 Views
Scene understanding in a contested battlefield is one of the very difficult tasks for detecting and identifying threats. In a complex battlefield, multiple autonomous robots for multi-domain operations are likely to track the activities of the same threat/objects leading to inefficient and redundant tasks. To address this problem, we propose a novel and effective object clustering framework that takes into account the position and depth of objects scattered in the scene. This framework enables the robot to focus solely on the objects of interest.
- Categories:
238 Views