The Video Segmentation Benchmark (VSB100) provides ground truth annotations for the Berkeley Video Dataset, which consists of 100 HD quality videos divided into train and test folders containing 40 and 60 videos, respectively. Each video was annotated by four different persons. The annotation comes with the evaluation software used for the experiments reported in Galasso et al. Download training set (40 Videos, external link to Berkeley server) Download Test set (60 Videos, external link to Berkeley server) Local links for Train and Test set Train set Test set Download annotation Full frame annotation - General Benchmark. Training set [Full Res] [Half Res] Test set [Full Res] [Half Res] Download evaluation code (ver 1.3) Denser annotations for training set by Khoreva et.al, GCPR 2014: [Full Res] [Half res] Subtasks Motion - objects which undergo significant motion. Objects such as snow, fire are excluded. Training set [Full Res] [Half Res] Test set [Full Res] [Half Res] Non-rigid motion - subset of Motion subtask, objects which undergo signficant articulated motion Training set [Full Res] [Half Res] Test set [Full Res] [Half Res] Camera motion - sequences where the camera moves Training set [Full Res] [Half Res] Test set [Full Res] [Half Res] Mirror links of VSB100 [Mirror 1] [Mirror 2] Terms of use The dataset is provided for research purposes. Any commercial use is prohibited. When using the dataset in your research work, you should cite the following papers: F. Galasso, N.S. Nagaraja, T.J. Cardenas, T. Brox, B. Schiele A Unified Video Segmentation Benchmark: Annotation, Metrics and Analysis, International Conference on Computer Vision (ICCV), 2013. P. Sundberg, T. Brox, M. Maire, P. Arbelaez, and J. Malik Occlusion Boundary Detection and Figure/Ground Assignment from Optical Flow, Computer Vision and Pattern Recognition (CVPR), 2011 .
The Freiburg-Berkeley Motion Segmentation Dataset (FBMS-59) is an extension of the BMS dataset with 33 additional video sequences. A total of 720 frames...
motion, benchmark, video, object, pedestrian, segmentation, tracking, groundtruthThe SegTrack dataset consists of six videos (five are used) with ground truth pixelwise segmentation (6th penguin is not usable). The dataset is used fo...
motion, video, object, proposal, flow, segmentation, stationary, model, camera, optical, groundtruthThe UCF Person and Car VideoSeg dataset consists of six videos with groundtruth for video object segmentation. Surfing, jumping, skiing, sliding, big ...
video, object, segmentation, motion, model, camera, groundtruthThe ICG Lab 6 (Multi-Camera Multi-Object Tracking) dataset contains 6 indoor people tracking scenarios recorded at our laboratory using 4 static Axis P1...
evaluation, graz, object, laboratory, pedestrian, segmentation, multiview, tracking, camera, detection, calibrationThe MOT Challenge is a framework for the fair evaluation of multiple people tracking algorithms. In this framework we provide: - A large collection of...
multiple, benchmark, evaluation, benhttp://motchallenge.net/chmark, dataset, target, video, pedestrian, 3d, tracking, surveillance, peopleThe UrbanStreet dataset used in the paper can be downloaded here [188M] . It contains 18 stereo sequences of pedestrians taken from a stereo rig mounted...
urban, human, recognition, video, pedestrian, segmentation, tracking, multitarget, detectionWe present the 2017 DAVIS Challenge, a public competition specifically designed for the task of video object segmentation. Following the footsteps of ot...
code, quality, benchmark, video segmentation, object, segmentation, hd, tracking, resolutionThe ICG Multi-Camera datasets consist of Easy Data Set (just one person) Medium Data Set (3-5 persons, used for the experiments) Hard Data Set (cro...
graz, indoor, video, object, pedestrian, multiview, tracking, camera, multitarget, detection, calibrationThe High Definition Analytics (HDA) dataset is a multi-camera High-Resolution image sequence dataset for research on High-Definition surveillance: Pedes...
high-definition, benchmark, human, lisbon, indoor, video, re-identification, pedestrian, network, multiview, tracking, surveillance, camera, detectionThe GaTech VideoSeg dataset consists of two (waterski and yunakim?) video sequences for object segmentation. There exists no groundtruth segmentation ...
video, object, segmentation, motion, model, cameraThe QMUL Junction dataset is a busy traffic scenario for research on activity analysis and behavior understanding. Video length: 1 hour (90000 frame...
video, motion, pedestrian, crowd, counting, tracking, detection, behaviorThe ICG Multi-Camera and Virtual PTZ dataset contains the video streams and calibrations of several static Axis P1347 cameras and one panoramic video fr...
graz, outdoor, video, object, panorama, pedestrian, network, crowd, multiview, tracking, camera, multitarget, detection, calibrationThe Berkeley Video Segmentation Dataset (BVSD) contains videos for segmentation (boundary?) Dataset train Dataset test
video, segmentation, benchmarkThe BEOID dataset includes object interactions ranging from preparing a coffee to operating a weight lifting machine and opening a door. The dataset is ...
video, object, egocentric, 3d, interaction, pose, trackingLASIESTA is composed by many real indoor and outdoor sequences organized in different categories, each of one covering a specific challenge in moving ob...
motion, subtraction, dataset, background, object, stationary, foreground, camera, challenge, detection, groundtruthMany different labeled video datasets have been collected over the past few years, but it is hard to compare them at a glance. So we have created a hand...
video, object, benchmark, classification, recognition, detection, actionThe VOT2016 pixel-wise annotations dataset contains pixel-wise per-frame annotations for sequences from VOT2016 dataset. The annotation is in a form of ...
object, segmentation, annotation, mask, visual, trackingSome datasets and evaluation tools are provided on this page for four different computer vision and computer graphics problems. Population counting L...
urban, surface, reconstruction, pointcloud, object, road, pedestrian, network, line, 3d, crowd, counting, detection, groundtruthThe Video Summarization (SumMe) dataset consists of 25 videos, each annotated with at least 15 human summaries (390 in total). The data consists of vide...
video, benchmark, summary, event, human, groundtruth, actionThe Mall dataset was collected from a publicly accessible webcam for crowd counting and profiling research. Ground truth: Over 60,000 pedestrians wer...
video, pedestrian, crowd, counting, tracking, detection, indoor, webcamThe Microsoft COCO (mscoco) is an image recognition and segmentation dataset which contains more 300k images for more than 70 categories. Other featur...
object, segmentation, benchmark, semantic, context, recognition, detectionCOCO-Stuff augments the COCO dataset with pixel-level stuff annotations for 10,000 images. These annotations can be used for scene understanding tasks l...
annotation, benchmark, coco, segmentation, things, captioning, stuff, groundtruth, semanticThe KTH Multiview Football dataset contains 771 images of football players includes images taken from 3 views at 257 time instances 14 annotated body jo...
recognition, soccer, outdoor, object, pedestrian, game, pose, multiview, tracking, camera, multitarget, detectionThe YouTube-Objects dataset is composed of videos collected from YouTube by querying for the names of 10 object classes. It contains between 9 and 24 vi...
video, object, flow, segmentation, detection, opticalThe PASCAL VOC is augmented with segmentation annotation for semantic parts of objects. For example, for the person category, we provide segmentation ma...
part, human, recognition, object, pedestrian, segmentation, pascal, detection, semanticPenn-Fudan Pedestrian Detection and Segmentation
segmentation, motion, background, pedestrian, detectionThe Airport MotionSeg dataset contains 12 sequences of videos of an aiprort scenario with small and large moving objects and various speeds. It is chall...
video, segmentation, motion, airport, clustering, camera, zoomBackground Models Challenge (BMC) is a complete dataset and competition for the comparison of background subtraction algorithms. The main topics concer...
motion, background, video, modeling, segmentation, change, surveillance, detectionWe present a new large-scale dataset that contains a diverse set of stereo video sequences recorded in street scenes from 50 different cities, with high...
urban, stereo, cities, person, video, weakly, segmentation, pedestrian, detection, car, semanticThe TUD Crossing dataset from Micha Andriluka, Stefan Roth and Bernt Schiele consists of 201 images with 1008 highly overlapping pedestrians with signif...
urban, sideview, overlap, segmentation, pedestrian, tracking, multitarget, detectionISPRS Test Project on Urban Classification, 3D Building Reconstruction and Semantic Labeling. In this part of our working group site you will get furthe...
urban, benchmark, recognition, aerial, canada, segmentation, photogrammetry, germany, 3d, multiview, city, semanticThe Lane Level Localization dataset was collected on a highway in San Francisco with the following properties: * Reasonable traffic * Multiple lane h...
driving, benchmark, autonomous, video, road, gps, map, 3d, localization, carThe Oxford RobotCar Dataset contains over 100 repetitions of a consistent route through Oxford, UK, captured over a period of over a year. The dataset c...
driving, street, urban, time, recognition, autonomous, video, segmentation, robot, classification, detection, car, yearThis web page contains video data and ground truth for 16 dances with two different dance patterns. The style of dancing is inspired by Scottish Ceilidh...
motion, dance, analysis, background, action, video, chemistry, patternThe SUNCG dataset is a Large 3D Model Repository for Indoor Scenes. SUNCG is an ongoing effort to establish a richly-annotated, large-scale dataset ...
scene, layout, recognition, indoor, object, segmentation, rendering, 3d, realism, room, syntheticThe video co-segmentation dataset contains 4 video sets which totally has 11 videos with 5 frames of each video labeled with the pixel-level ground-tr...
video, segmentation, co-segmentation, datasetThe Multicamera Human Action Video Data (MuHAVi) Manually Annotated Silhouette Data (MAS) are two datasets consisting of selected action sequences for ...
video, segmentation, action, behavior, human, backgroundAt Udacity, we believe in democratizing education. How can we provide opportunity to everyone on the planet? We also believe in teaching really amazing ...
driving, street, urban, time, recognition, autonomous, video, segmentation, robot, classification, detection, car, syntheticThe Leeds Cows dataset by Derek Magee consists of 14 different video sequences showing a total of 18 cows walking from right to left in front of differe...
video, segmentation, detection, cow, animal, backgroundThe Daimler Mono Pedestrian Classification Benchmark dataset consists of two parts: a base data set. The base data set contains a total of 4000 pedest...
illumination, object, urban, pedestrian, classification, outdoor, scaleThe TUD Pedestrians dataset from Micha Andriluka, Stefan Roth and Bernt Schiele [AndrilukaCVPR2008] consists of 250 images with 311 fully visible people...
segmentation, pedestrian, sideview, object detectionJPL First-Person Interaction dataset (JPL-Interaction dataset) is composed of human activity videos taken from a first-person viewpoint. The dataset par...
video, motion, action, interactive, recognition, humanAbstract Scene understanding has (again) become a focus of computer vision research, leveraging advances in detection, context modeling, and tracking. ...
scene, segmentation, pedestrian, 3d, classification, understanding, car, semanticThe Outex dataset is part of a framework for empirical evaluation of texture classification and segmentation algorithms. The framework is being constru...
segmentation, benchmark, classification, synthetic, textureThe MSR RGB-D Dataset 7-Scenes dataset is a collection of tracked RGB-D camera frames. The dataset may be used for evaluation of methods for different a...
video, kinect, location, reconstruction, depth, trackingThe Fish4Knowledge project (groups.inf.ed.ac.uk/f4k/) is pleased to announce the availability of 2 subsets of our tropical coral reef fish video and e...
motion, nature, recognition, fish, video, water, classification, animal, cameraThe GaTech VideoContext dataset consists of over 100 groundtruth annotated outdoor videos with over 20000 frames for the task of geometric context eval...
urban, nature, outdoor, video, segmentation, supervised, classification, context, unsupervised, geometry, semanticThis ETHZ CVL RueMonge 2014 dataset used for 3D reconstruction and semantic mesh labelling for urban scene understanding. It was first published in [1...
benchmark, paris, reconstruction, pointcloud, outdoor, 3d, source, architecture, semantic, code, urban, mesh, recognition, segmentation, classificationThe test sequences provide interested researchers a real-world multi-view test data set captured in the blue-c portals. The data is meant to be used for...
tracking, segmentation, camera, action, multiviewThe Pedestrian Parsing dataset contains 3,673 images from 171 videos of different Surveillance Scenes (PPSS), where 2,064 images are occluded and 1,609 ...
segmentation, pedestrian, parsingHollywood-2 datset contains 12 classes of human actions and 10 classes of scenes distributed over 3669 video clips and approximately 20.1 hours of video...
video, segmentation, action classificationt is composed of food intake movements, recorded with Kinect V1 (320240 depth frame resolution), simulated by 35 volunteers for a total of 48 tests. The...
kinect, age, intake, pointcloud, human, tracking, monitoring, groundtruth, food, behaviorMICCAI 2015 Challenge on Liver Ultrasound Tracking Munich, October 9, 2015 (Full Day) Outline Ultrasound (US) imaging is a widely used medical imagi...
ultrasound, liver, benchmark, real, therapy, human, medical, tracking, organ3 datasets: PTZ Tracking, Thermal-visible registration, Single object tracking
tracking, pedestrian, thermal, ptzGaze data on video stimuli for computer vision and visual analytics. Converted 318 video sequences from several different gaze tracking data sets with...
video, metadata, segmentation, gaze data, polygon annotationThe dataset contains 15 documentary films that are downloaded from YouTube, whose durations vary from 9 minutes to as long as 50 minutes, and the total ...
video, object, detectionRobust Multi-Person Tracking from Mobile Platforms In all cases, data was recorded using a pair of AVT Marlins F033C mounted on a chariot respectively...
tracking, pedestrian, color, sequenceThe Brodatz dataset consists of 112 textures in grayscale images of various texture types. http://www.ee.oulu.fi/research/imag/texture/image_data/Brod...
segmentation, benchmark, classification, synthetic, textureThe 1DSfM Landmarks is a collection of community-based image reconstruction by Kyle Wilson and is comprised of 14 datasets with comparison to bundler gr...
urban, 3d, benchmark, city, reconstruction, landmark, groundtruthThe TUD Campus dataset from Micha Andriluka, Stefan Roth and Bernt Schiele consists of 71 images and 303 highly overlapping pedestrians with large scale...
segmentation, pedestrian, sideview, object tracking, object detection, overlapScanNet is an RGB-D video dataset containing 2.5 million views in more than 1500 scans, annotated with 3D camera poses, surface reconstructions, and ins...
scene, layout, recognition, indoor, object, cad, segmentation, rendering, 3d, realism, room, syntheticThe SPHERE human skeleton movements dataset was created using a Kinect camera, that measures distances and provides a depth map of the scene instead of ...
motion, skeleton, kinect, movement, depth, human, action, video, behaviorPedestrian Color Naming (PCN) dataset contains 14,213 images, each of which hand-labeled with color label for each pixel.
segmentation, pedestrian, color namingThe Daimler Mono Pedestrian Detection Benchmark dataset contains a large training and test set. The training set contains 15.560 pedestrian samples (ima...
object, mono, urban, pedestrian, outdoor, scale, detectionScene Parsing Benchmark Scene parsing data and part segmentation data derived from ADE20K dataset could be download from MIT Scene Parsing Benchmark. ...
segmentation, annotation, benchmark, semantic, scene, recognitionThe HandNet dataset contains depth images of 10 participants hands non-rigidly deforming infront of a RealSense RGB-D camera. This dataset includes 2...
rgbd, hand, articulation, video, segmentation, classification, pose, fingertip, detectionThe dataset captures 25 people preparing 2 mixed salads each and contains over 4h of annotated accelerometer and RGB-D video data. Annotated activities ...
video, activity, classification, tracking, recognition, detection, actionThe Traffic Video dataset consists of X video of an overhead camera showing a street crossing with multiple traffic scenarios. The dataset can be down...
video, urban, traffic, road, overhead, tracking, view, detectionThe current video database containing six types of human actions (walking, jogging, running, boxing, hand waving and hand clapping) performed several ti...
video, segmentation, action classificationThe Weizmann actions dataset by Blank, Gorelick, Shechtman, Irani, and Basri consists of ten different types of actions: bending, jumping jack, jumping,...
video, segmentation, action, action classificationThe .enpeda.. Image Sequence Analysis Test Site (EISATS) offers sets of long bi- or trinocular image sequences recorded in the context of vision-based d...
motion, stereo, analysis, flow, segmentation, optical, semantic, visionThe PETS 2009 dataset contains 3 parts showing multi-view sequences containing pedestrians walking in an outdoor environment. The parts are used for per...
overlap, human, frontview, occlusion multitarget, outdoor, pedestrian, tracking, detectionThe Inria Aerial Image Labeling addresses a core topic in remote sensing: the automatic pixelwise labeling of aerial imagery (link to paper). Dataset ...
house, urban, aerial, building, segmentation, footprint, groundtruth, city, semanticThe TU Berlin Multi-Object and Multi-Camera Tracking Dataset (MOCAT) is a synthetic dataset to train and test tracking and detection systems in a virtua...
evaluation, multi-view, pedestrian, animal, tracking, multi-class, vehicle, detection, syntheticThe UMD Dynamic Scene Recognition dataset consists of 13 classes and 10 videos per class and is used to classify dynamic scenes. The dataset has been ...
video, motion, dynamic, classification, scene, recognitionWelcome to the homepage of the gvvperfcapeva datasets. This site serves as a hub to access a wide range of datasets that have been created for projects ...
face, reconstruction, depth, mesh, human, action, video, pose, multiview, trackingWe would like to announce the release of PASCAL-Context dataset. We augmented PASCAL VOC 2010 dataset with annotations for 400+ additional categories. I...
segmentation, benchmark, shape, recognition, pascal, category, semantic, denseThe crowd datasets are collected from a variety of sources, such as UCF and data-driven crowd datasets. The sequences are diverse, representing dense cr...
video, pedestrian, scene, crowd, human, understanding, anomaly, detectionThe Textures volume currently contains 154 images, all monochrome, 129 512x512 and 25 1024x1024. For the Brodatz texture images, the number in parenth...
segmentation, benchmark, evaluation, classification, synthetic, textureThe Where Who Why (WWW) dataset provides 10,000 videos with over 8 million frames from 8,257 diverse scenes, therefore offering a superior comprehensive...
recognition, video, flow, pedestrian, crowd, surveillance, optical, detectionThe CVC Partial Occlusion Virtual Pedestrian datasets (CVC-01 to CVC-06) cover a range of scenarios of occluded pedestrians generated in a virtual and r...
urban, pedestrian, classification, synthetic, occlusion, tracking, detectionThe Yotta dataset consists of 70 images for semantic labeling given in 11 classes. It also contains multiple videos and camera matrices for 14km or driv...
urban, reconstruction, video, segmentation, 3d, classification, camera, semanticThe multiple foreground video co-segmentation dataset, consisting of four sets, each with a video pair and two foreground objects in common. The datase...
video, segmentation, co-segmentationThe Daimler Urban Segmentation Dataset consists of video sequences recorded in urban traffic. The dataset consists of 5000 rectified stereo image pairs ...
segmentation, urban, motion, stereo, semantic, outdoorThe TUD Pedestrians training dataset from Micha Andriluka, Stefan Roth and Bernt Schiele consists of 210 and 400 training images with X pedestrians with...
segmentation, pedestrian, sideview, object detectionThe Farman Institute 3D Point Sets dataset contains 11 objects by a 3D laser scanner. This dataset was peer-reviewed by Image Processing On Line: Farman...
object, scanner, 3d, reconstruction, point, model, laserThe CERTH image blur dataset consists of 2450 digital images, 1850 out of which are photographs captured by various camera models in different shooting ...
motion, quality, detection, image, defocus, blurAn indoor action recognition dataset which consists of 18 classes performed by 20 individuals. Each action is individually performed for 8 times (4 dayt...
video, open-view, cross-view, recognition, indoor, action, multi-cameraISPRS / EuroSDR Benchmark for Multi-Platform Photogrammetry In these pages you can get information about the BENCHMARK FOR MULTI-PLATFORM PHOTOGRAMMET...
urban, benchmark, reconstruction, aerial, photogrammetry, germany, 3d, multiview, switzerland, cityThis dataset package contains the software and data used for Detection-based Object Labeling on the RGB-D Scenes Dataset as implemented in the paper: ...
object, 3d, kinect, reconstruction, depth, recognition, indoorThe INRIA Horses dataset from Frederic Jurie and Vittorio Ferrari consists of 170 images with one or more horses in side-view at several scales and clut...
segmentation, clutter, horse, object detection by shape, natureThe database of nude and non-nude videos contains a collection of 179 video segments collected from the following movies: Alpha Dog, Basic Instinct, Bef...
video, nude detection, movieWe take advantage of our autonomous driving platform Annieway to develop novel challenging real-world computer vision benchmarks. Our tasks of interest ...
stereo, object tracking, depth, reconstruction, detection tracking, object detection, segmentation, odometry, optical flow, semantic car depth, sfmThe object is a plaster reproduction of Temple of the Dioskouroi in Agrigento, Sicily. Click on thumbnail for a full-sized (640x480) image. Resolution o...
3d reconstruction, 3d, benchmark, sfm, multiviewThe dataset consist of the about 50 hours obtained from kindergarten surveillance videos. Dataset, totally approximately 100 videos sequences (1000GB, 5...
segmentation, action, behavior, video surveillance, human, backgroundMIT Pedestrian dataset from Papageorgiou and Poggio [IJCV2000] contains 509 training and 200 test images of pedestrians in city scenes (plus left-right ...
urban, pedestrian, boundingbox, frontview, people, object detectionThe 3D shape description dataset consists of multiple sub-datasets Descriptor Matching - Dataset 1 & 2 (Stanford) These datasets, created from some o...
description, 3d, benchmark, registration, reconstruction, shape, matchingThe MSR Action datasets is a collection of various 3D datasets for action recognition. See details http://research.microsoft.com/en-us/um/people/zliu...
video, detection, 3d, action, reconstruction, recognitionThe PASCAL VOC Challenge datasets by Mark Everingham is a yearly dataset which has a central evaluation server and the final test data is not released. ...
chair, object detection, building, object segmentation, pedestrian, object pose, animal, car, airplaneUnlike the previous SHREC contests, the objective of this SHREC 2012 contest is to evaluate the performance of 3D-mesh segmentation techniques instead o...
mesh, segmentation, 3d, partThe Prague Texture Segmentation Datagenerator and Benchmark is designed to mutually compare and rank different (dynamic/static) texture segmenters (supe...
benchmark, texture segmentation, texture classification, syntheticPermanently growing database on lung tuberculosis patients. The data include radiological images (CT+XRay) plus social, clinical, and lab data as well a...
medical, segmentation, xray, chest, genome, none, tuberculosis, ctThe TUG (Timed Up and Go test) dataset consists of actions performed three times by 20 volunteers. The people involved in the test are aged between 22 a...
wearable, kinect, time, human, recognition, action, depth image processing - tug, accelerometer, videoThe procedural texture perceptual similarity dataset contains a list of procedural textures along with their pairwise distances, as defined by a percept...
study, benchmark, procedural, textureThis dataset is for people tracking in wide baseline camera networks and was designed as a contest at ICPR 2012. The contest consists of two challeng...
trajectory, pedestrian, occlusion, crowd, object tracking, overlap, aerial, object detectionThe Weather and Illumination Database (WILD) is an extensive database of high quality images of an outdoor urban scene, acquired every hour over all sea...
urban, estimation, depth, weather, time, newyork, webcam, video, illumination, change, static, camera, lightSceneNet RGB-D is dataset comprised of 5 million Photorealistic Images of Synthetic Indoor Trajectories with Ground Truth. It expands the previous work ...
trajectory, reconstruction, scene, slam, lighting, indoor, segmentation, robot, rendering, 3d, synthetic, navigationThe Visual Attributes dataset contains visual attribute annotations for over 500 object classes (animate and inanimate) which are all represented in Ima...
object, recognition, attribute, classification, imagenetThis dataset contains videos of crowds and other high density moving objects. The videos are collected mainly from the BBC Motion Gallery and Getty Imag...
segmentationThe 3D Mask Attack Database (3DMAD) is a biometric (face) spoofing database. It currently contains 76500 frames of 17 persons, recorded using Kinect for...
face, emotion, segmentation, 3d, recognition, biometry, frontviewThe Salient Montages is a human-centric video summarization dataset from the paper [1]. In [1], we present a novel method to generate salient montages...
video, saliency, wearable, montage, summarization, humanThe Longterm Pedestrian dataset consists of images from a stationary camera running 24 hours for 7 days at about 1 fps. It used for adaptive detection ...
coffee, graz, background, indoor, illumination, change, pedestrian, robust, multitarget, detectionISPRS and EuroSDR - Benchmark on High Density Aerial Image Matching Background and Scope of the project Innovations in matching algorithms as well a...
urban, benchmark, reconstruction, aerial, photogrammetry, germany, 3d, multiview, switzerland, cityThe xawAR16 dataset is a multi-RGBD camera dataset, generated inside an operating room (IHU Strasbourg), which was designed to evaluate tracking/relocal...
video, medicine, table, depth, operation, recognition, surgeryThe Pornography database contains nearly 80 hours of 400 pornographic and 400 non-pornographic videos. For the pornographic class, we have browsed websi...
video, pornography, video shots, video framesThe domain-specific personal videos highlight dataset from the paper [1] describes a fully automatic method to train domain-specific highlight ranker f...
saliency, domain, wearable, human, recognition, action, video, summarizationThe UK Bench dataset from Henrik Stewenius and David Nister contains 10200 images of N=2550 groups with each four images at size 640x480. The images are...
image retrieval, object, rotation, centeredLow-resolution RGB videos + ground truth trajectories from multiple fixed and moving cameras monitoring the same scenes (indoor and outdoor) to improve ...
trackingShakeFive2 A collection of 8 dyadic human interactions with accompanying skeleton metadata. The metadata is frame based xml data containing the skelet...
video, human, kinect, interactionThis dataset contains 7 challenging volleyball activity classes annotated in 6 videos from professionals in the Austrian Volley League (season 2011/12)....
video, sport, analysis, activity recognition, volleyball, detection, action10 videos as inputs, and segmented image sequences as ground-truth
segmentationDataset contains 1000 images of 100 persons, with 10 images per person and is freely available. All images were acquired by cropping ears from images fr...
person, pedestrian, ear, recognition, human, lighting, biometryThe Multi-illuminant Image Sequences dataset contains 16 video sequences (13 with single light source and 3 with two global light sources), recorded wi...
constancy, color, white, chromaticity, physics, nature, dichromatic, illumination, object, balance, lightAn Annotated Dataset For Near-Duplicate Detection In Personal Photo Collections Managing photo collections involves a variety of image quality assessm...
copyright, duplicate, detection, groundtruth, retrievalThe PETS 2016 IPATCH dataset contains a set of fourteen multi camera recordings (visible, themal) collected off the coast of Brest, France, in collabora...
visible, thermal, multimodal, vessel, maritime, boat, gps, tracking, detection, radarCows for object segmentation, Five video sequences for motion segmentation
segmentationA 66 stereo pairs dataset with their subpixel ground truths. The construction and improvement of algorithms for subpixel stereovision requires very pr...
stereo, depth, pointcloud, noise, stereovision, 3d, groundtruth, subpixelDaimler Stereo Pedestrian Detection Benchmark C. Keller, M. Enzweiler, and D. M. Gavrila, A New Benchmark for Stereo-based Pedestrian Detection, Proc...
urban, pedestrian, object detectionHallway Corridor - Multiple Camera Tracking: An indoor camera network dataset with 6 cameras (contains ground plane homography).
trackingDaimler Multi-Cue, Occluded Pedestrian Classification Benchmark Training and test samples have a resolution of 48 x 96 pixels with a 12-pixel border a...
image classification, urban, pedestrian, object detectionWe provide the three datasets used for testing our system for our ICCV 2007 publication, including annotations. Data was recorded using a pair of AVT Ma...
trajectory, pedestrian, occlusion, crowd, object tracking, object detection, overlapThe Semantic Description of Human Activities (SDHA) was a contest at ICPR 2010. The contest is composed of three different types of activity recognit...
trajectory, pedestrian, occlusion, crowd, object tracking, overlap, aerial, object detectionThe tracking environment consists of multiple 3D range sensors, covering an area of about 900 m2, in the "ATC" shopping center in Osaka, Japan.
trackingFor the first few decades of the fields existence, computer vision has been focused on algorithmic, logical approaches to perception. But it was only wi...
object, 3d, kinect, reconstruction, depth, recognition, indoorThe automotive multi-sensor (AMUSE) dataset consists of inertial and other complementary sensor data combined with monocular, omnidirectional, high fram...
urban, api, image, video, inertial, streetside, traffic, cityThe Zurich Summer v1.0 dataset is a collection of 20 chips (crops), taken from a QuickBird acquisition of the city of Zurich (Switzerland) in August 200...
annotation, urban, pan, gsd, superpixel, nir, aerial, satellite, segmentation, zurich, rgb, city, semanticImage segmentation and boundary detection. Grayscale and color segmentations for 300 images, the images are divided into a training set of 200 images, a...
segmentationPN Learning - How does TLD work? Tracking estimates the object location as long as the object is visible. During tracking all observed patterns of the...
face, pedestrian, singletarget, learning, bike, object trackingAWS hosts a variety of public datasets that anyone can access for free. Previously, large datasets such as satellite imagery or genomic data have requi...
space, human, recognition, image, amazon, satellite, segmentation, learning, deep, classification, biology, resolutionMultispectral Imaging (MSI) datasets were acquired using IRIS II which is a lightweight portable system comprising of a high resolution camera, a novel ...
illumination, wavelength, registration, alignment, matching, groundtruth, multi-spectralThe CIFAR-10 dataset consists of 60000 32x32 colour images in 10 classes, with 6000 images per class. There are 50000 training images and 10000 test ima...
object, color, patch, scene, tiny, image classificationThe INRIA People dataset from Navneet Dalal and Bill Triggs [DalalCVPR2005] consists of training and testing data. The training contains 1805 images and...
pedestrian, sideview, boundingbox, frontview, object detection, humanThe set was recorded in Zurich, using a pair of cameras mounted on a mobile platform. It contains 12'298 annotated pedestrians in roughly 2'000 frames.
trackingCollected in a clothing store. Captured with Kinect (640*480, about 30fps)
tracking, detectionContains hand-labelled pixel annotations for 38 groups of images, each group containing a common foreground. Approximately 17 images per group, 643 imag...
segmentationThe mirror symmetry database contains 176 single-symmetry and 63 multyple-symmetry images (.png files) with accompanying ground-truth annotations (.mat ...
symmetry, detection, groundtruth, mirrorA new large-scale PEdesTrian Attribute (PETA) dataset. The dataset is by far the largest of its kind, covering more than 60 attributes on 19000 images. ...
pedestrian, crowdThe Buffy dataset contains images selected from the TV series, Buffy: the Vampire Slayer. We select a set of 452 images from the first two episodes for ...
segmentation, human, buffy, movie, object detectionThe dataset is composed of 150 synthetic scenes, captured with a (perspective) virtual camera, and each scene contains 3 to 5 objects. The model set is ...
mesh, segmentation, recognition, syntheticThe multi-modal/multi-view datasets are created in a cooperation between University of Surrey and Double Negative within the EU FP7 IMPART project. Th...
rgbd, color, dynamic, multi-view, action, outdoor, video, 3d, face, emotion, lidar, human, indoor, multi-mode, modelThe THUS10000 benchmark dataset comprises of 10,000 images, each of which has an unambiguous salient object and the object region is accurately annotate...
saliency, segmentation, salient object detection, attention, visualThe database contains, for each of the 100 examples: (1) the uncompressed frames, up to the 10th frame after the appearance of the 8th cell; (2) a text ...
trajectory, circle, mouse, biology, cell, trackingThe Caltech Pedestrian Dataset consists of approximately 10 hours of 640x480 30Hz video taken from a vehicle driving through regular traffic in an urban...
urban, pedestrian, object detectionThe 2D-3D-S dataset provides a variety of mutually registered modalities from 2D, 2.5D and 3D domains, with instance-level semantic and geometric annota...
reconstruction, depth, large-scale, indoor, normal, building, panorama, segmentation, 3d, semanticScene Background Initialization (SBI) dataset The SBI dataset has been assembled in order to evaluate and compare the results of background initializa...
change, detection, benchmark, background, foreground, initializationThe Paris Art Deco Facades dataset consists of 79 / 80 images of rectified facades of the architectural style Art Deco, which has different sizes of win...
urban, paris, grammar, facade, recognition, segmentation, procedural, architecture, semantic, cityThe Cholec80 dataset contains 80 videos of cholecystectomy surgeries performed by 13 surgeons. The videos are captured at 25 fps. The dataset is labeled...
video, medicine, surgery, phase, tool, recognitionIn HouseCraft, we utilize rental ads to create realistic textured 3D models of building exteriors. In particular, we exploit the address of the property...
house, urban, registration, floorplan, building, streetview, segmentation, localization, city, semanticDatabase contains 798 images of 114 persons, with 7 images per person and is freely available for research purposes. All images were taken in supervised...
face, person, human, lighting, recognition, illumination, pedestrian, biometryThe All I Have Seen (AIHS) dataset is created to study the properties of total visual input in humans, for around two weeks Nebojsa Jojic wore a camera ...
similarity, scene, summary, user, indoor, outdoor, video, 3d, clustering, studyIdiap/ETHZ Faces and Poses Dataset dataset by L. Jie, B. Caputo and V. Ferrari contains 1703 image-caption pairs. [author] Captions contain the names of...
face detection, pedestrian, text, object poseThis data set comprises 144 images of an edge profile cutting head of a milling machine. The head tool contains a total of 30 cutting inserts. The cutti...
profile, head, cutting, edge, tools, inserts, object, tool, milling, localization, wear, monitoringA dataset acquired with 3 synchronized sensors (Primesense Carmine 1.09, Microsoft Kinect v2, Canon IXUS 950 IS), featuring: * 30 industry-relevant ob...
object, rgbd, 3d, estimation, pose, texture-lessThe ETHZ Shape classes dataset from Vittorio Ferrari [?] consists of five object classes and a total of 255 images. All classes contain significant intr...
clutter, swan, bottle, matching, nature, object detection by shape, mug, giraffe, segmentation, applelogoThe dataset consists of eight unique scenes in crowded spaces such as a university campus or the sidewalks of a busy street.
trackingThe TRaffic ANd COngestionS (TRANCOS) dataset, a novel benchmark for (extremely overlapping) vehicle counting in traffic congestion situations. It consi...
urban, highway, spain, object, traffic, transportation, vehicle, detection, carIt is composed of ADL (activity daily living) and fall actions simulated by 11 volunteers. The people involved in the test are aged between 22 and 39, w...
wearable, kinect, fall detection - adl, depth, human, recognition, action, accelerometer, videoThis work attempts to provide two Hand Images Databases for hand biometrics: one is created using a mobile phone camera of modest quality, which we ca...
segmentation, person, identification, authentication, mobile, shape, biometric/ hand geometry, webcamThe Multi-FoV synthetic datasets are two synthetic scenes (vehicle moving in a city, and flying robot hovering in a confined room). For each scene, thre...
synthetic, visual, odometry, fov, blender, camera, groundtruthThis dataset consists of more than 22,000 images of 24 people which are captured by 16 cameras installed in a shopping mall "Shinpuh-kan". All images ar...
trackingMIT traffic data set is for research on activity analysis and crowded scenes. It includes a traffic video sequence of 90 minutes long. It is recorded by...
trackingThe Ford Car dataset is joint effort of Pandey et al. (for collecting images, Lidar points, calibration etc.) and us (for annotation of 2D and 3D object...
lidar, detection, groundtruth, 3d, car, sfmThe DynTex dataset consists of a comprehensive set of Dynamic Textures. Dynamic, or temporal, texture is a spatially repetitive, time-varying visual pat...
segmentation, dynamic, video repetition, synthetic, textureThe Babenko tracking dataset contains 12 video sequences for single object tracking. For each clip they provide (1) a directory with the original i...
face, video, single, occlusion, object tracking, animalClassification/Detection Competitions, Segmentation Competition, Person Layout Taster Competition datasets
segmentationThe Leuven Stereo Scene dataset is a scene and depth dataset. There exist two variants of this dataset - a CVPR 2007 paper [1] by Leibe et al. for detec...
urban, stereo, depth, reconstruction, leuven, segmentation, 3d, semantic, sfmThe Extreme Classification Repository: Multi-label Datasets & Code Kush Bhatia Himanshu Jain Prateek Jain Manik Varma The objective in extreme mu...
multilabel, machine, learning, benchmark, evaluation, classificationGround truth database of 50 images with: Data, Segmentation, Labelling - Lasso, Labelling - Rectangle
segmentationThe Graz01 dataset by Andreas Opelt and Axel Pinz contains four types of images: bikes, people, background with no bikes, background with no people.
pedestrian, clutter, occlusion, object detection, graz, background, bikeGeometric Context Dataset: pixel labels for seven geometric classes for 300 images
segmentationThe VSUMM (Video SUMMarization) dataset is of 50 videos from Open Video. All videos are in MPEG-1 format (30 fps, 352 x 240 pixels), in color and with s...
similarity, type, summary, user, video, static, keyframe, studyWe developed imaging hardware consisting of a color camera, a thermal camera and a beam splitter to capture the aligned multispectral (RGB color + Therm...
pedestrian, thermal, rgbFine-Grained Visual Classification of Aircraft (FGVC-Aircraft) is a benchmark dataset for the fine grained visual categorization of aircraft. Data, an...
benchmark, evaluation, fine-grained, classification, aircraft, airplane, recognitionWe wanted to have a collection of action recognition papers and results that everybody can use for reference. The site will work by the community princi...
recognition, benchmark, action, datasetThe LabelMeFacade dataset contains buildings, windows, sky and a limited number of unlabeled regions (maximally 20% covering of the image). This procedu...
segmentation, urban, semantic, recognition, facade, rectifiedThe Aspect Layout dataset is designed to allow evaluation of object detection for aspect ratios in perspective images. Author text: In this project ...
object, detection, aspect, perspective, ratio, layoutThe Geosemantic is a dataset of object locations from GIS and a query image with metadata. It is used to project the buildings and streets that are in t...
geography, gps, segmentation, gis, supervised, semanticThe Malaya Abrupt Motion (MAMo) dataset is targeted for visual tracking, particularly for abrupt motion tracking. It was collected from publicly accessi...
abrupt motion tracking, tracking, visual trackingThe Graz02 dataset by Andreas Opelt and Axel Pinz contains four categories of images: bikes, people, cars and a single background class. The annotation ...
pedestrian, clutter, object detection, graz, background, car, bikeParis-rue-Madame dataset contains 3D Mobile Laser Scanning (MLS) data from rue Madame, a street in the 6th Parisian district (France). The test zone con...
segmentation, 3d, semantic, classification, pointcloud, laserThe CHALEARN Multi-modal Gesture Challenge is a dataset +700 sequences for gesture recognition using images, kinect depth, segmentation and skeleton dat...
gesture, skeleton, kinect, depth, human, recognition, action, illumination, segmentationThis is a subset of the dataset introduced in the SIGGRAPH Asia 2009 paper, Webcam Clip Art: Appearance and Illuminant Transfer from Time-lapse Sequence...
urban, nature, time, webcam, video, illumination, change, static, camera, lightThe contour patches dataset is a large dataset of images patch matches used for contour detection. References: C. L. Zitnick and D. Parikh The Role...
lowlevel, match, edge, image, contour, segmentation, patch, detectionThe Stanford Background Dataset is a new dataset introduced in Gould et al. (ICCV 2009) for evaluating methods for geometric and semantic scene understa...
segmentation, urban, geometry, semantic, classification, natureThe We Are Family Stickmen dataset from Eichner and Ferrari contains X images with X people in group photos for human pose estimation with annotated 2D ...
body part, pedestrian, object poseThe Comprehensive Cars (CompCars) dataset contains data from two scenarios, including images from web-nature and surveillance-nature. The web-nature dat...
object, urban, fine-grained, classification, recognition, vehicle, car, attributeThe Quad 6K dataset is a Structure-from-Motion dataset taken at Arts Quad at Cornell University campus and consists of 6514 images with ground truth pos...
urban, 3d reconstruction, groundtruth, sfm, landmark, 3d gpsThe Graffiti dataset by Krystian Mikolajczyk and Cordelia Schmid contains 48 images split into 8 sequences with 6 images each showing different structur...
benchmark, image rectification, feature detection, feature descriptionDataset A (former NLPR Gait Database) was created on Dec. 10, 2001, including 20 persons. Each person has 12 image sequences, 4 sequences for each of th...
motion, foot, human, recognition, gait, action, classification, biometry, pressureThe PSU HUB dataset is a detection, tracking dataset. Ground truth trajectory and grouping information for pedestrians walking in the PSU student union ...
trajectory, pedestrian, occlusion, crowd, object tracking, object detection, overlapThe Yale Face dataset from A. Georghiades contains 5760 single light source images of ten subjects, each shown in 9 poses and 64 illumination setups (le...
illumination, face detection, pedestrian, pose estimationThese datasets were generated for the M2CAI challenges, a satellite event of MICCAI 2016 in Athens. Two datasets are available for two different challen...
video, medicine, workflow, surgery, recognition, challengeThe object is a plaster dinosaur (stegosaurus). Click on thumbnail for a full-sized (640x480) image. Resolution of ground truth model: 0.00025m (you may...
3d reconstruction, 3d, benchmark, sfm, multiviewThe Pittsburgh Fast-food Image dataset (PFID) consists of 4545 still images, 606 stereo pairs, 3033600 videos for structure from motion, and 27 privacy-...
video, laboratory, classification, reconstruction, real, food, recognitionAn evaluation benchmark for dense MVS for these datasets fountain-P11, Herz-Jesu-P8, entry-P10, castle-P19, Herz-Jesu-P25, castle-P30 . Images (correc...
3d reconstruction, benchmark, sfm, depth, dense, meshThe CALTECH 256 dataset by Li Fei-Fei contains 30607 images for 256 categories.
object, detection, image, centered, classification, sceneThe HCI 4D Lightfields dataset contains 11 objects with corresponding lightfields for depth estimation. Datasets can be downloaded individually below....
3d, benchmark, evaluation, reconstruction, depth, 4d, lightfieldThe CMP Facade dataset consists of facade images assembled at the Center for Machine Perception, which includes 600 rectified images of facades from var...
urban, similarity, facade, recognition, segmentation, structure, classification, rectification, semanticThe Webcam Interestingness dataset consists of 20 different webcam streams, with 159 images each. It is annotated with interestingness ground truth, acq...
video, interest, retrieval, classification, weather, ranking, webcamThe TVPR dataset includes 23 registration sessions. Each of the 23 folders contains the video of one registration session. Acquisitions have been perfor...
person, depth, recognition, indoor, top-view, video, clothing, gender, reidentification, identification, peopleThese sequences were used for our video interpolation work described in High-quality video view interpolation using a layered representation, C.L. Zi...
segmentation, 3d reconstruction, camera, depthThis dataset consist 51 oral presentation recorded with 2 ambient visual sensor (web-cam), 3 First Person View (FPV) cameras (1 on presenter and 2 on ra...
video, quality, kinect, multi-sensor, presentation, analysis328 side-view color images of horses that were manually segmented. The images were randomly collected from the WWW.
segmentationThe VidPairs dataset contains 133 pairs of images, taken from 1080p HD (~2 megapixel) official movie trailers. Each pair consists of images of the same ...
matching, dense, video, flow, description, patch, pair, opticalThe CALTECH 101 dataset by Li Fei-Fei contains images for 101 categories with about 40 to 800 images per category. Most categories have about 50 images ...
object, natural-image, centered, scene, image classificationThe GaTech VideoStab dataset consists of N videos for the task of video stabilization. This code is implemented in Youtube video editor for stabilizatio...
video, camera, path, stabilizationThis material is supplementary to Michael Stark, Bernt Schiele. How Good are Local Features for Classes of Geometric Objects. Eleventh IEEE Internat...
object, binary, tool, classification, shapeThe PETS 2006 dataset contains 7 parts showing multi-sensor sequences containing left-luggage scenarios with increasing scene complexity at a train stat...
pedestrian, indoor, frontview, object tracking, object detection, multitargetUBC3V is a synthetic dataset for training and evaluation of single or multiview depth-based pose estimation techniques. The nature of the data is simila...
segmentation, multiview depth based pose estimationThe Person Re-ID (PRID) 2011 dataset was created in co-operation with the Austrian Institute of Technology for the purpose of testing person re-identifi...
trajectory, graz, illumination, pedestrian, change, appearance, identification, classification, multiview200 gray level images along with ground truth segmentations
segmentationTo encourage the open comparison of single image shadow removal in community, we provide an online benchmark site and a dataset. Our quantitatively veri...
illumination, shadow, benchmark, singleview, removalThe Western GCO Segmentation problem instances are provided to compare effects of graph size, neighborhood size, length of s to t paths, regional arc co...
face, adhead, abdomen, liver, binary, medical, segmentation, optimization, bone, babyfaceThe ETHZ Extended Shape classes dataset from Konrad Schindler is larger dataset of shape categories, created by merging ETHZ shape classes with Konrad S...
segmentation, clutter, object detection by shapeA New Color Image Database for Benchmarking of Face Detection Techniques and Human Skin Segmentation Techniques. A new color face image database for ...
face, segmentation, skin, detection, benchmarkingThe RGB-D Person Re-identification dataset is for person re-identification using depth information. The main motivation is that the standard techniques ...
pedestrian, 3d, identification, classification, depth, shapeWe introduce the Shelf dataset for multiple human pose estimation from multiple views. In addition we annotate the body joints in the Campus dataset fro...
motion, multiple, 3d, estimation, capture, pose, human, viewWe collected a video dataset, termed ChokePoint, designed for experiments in person identification/verification under real-world surveillance conditions...
face, real, human, recognition, world, pedestrian, identification, clustering, multiview, surveillance, detection, sequenceChairGest is an open challenge / benchmark. The task consists in spotting and recognizing gestures from multiple synchronized sensors: 1 Kinect and 4 X...
gesture, detection, benchmark, kinect, recognition, humanThe Berkeley Multimodal Human Action Database (MHAD) contains 11 actions performed by 7 male and 5 female subjects in the range 23-30 years of age excep...
recognition, motion, action, classification, multiviewThe FAce Semantic SEGmentation (FASSEG) repository contains datasets for multi-class semantic face segmentation. The FASSEG repository is composed by ...
face, segmentationThe BIWI Walking Pedestrians (EWAP) dataset shows walking pedestrians in busy scenarios from a bird eye view. Manually annotated. Data used for training...
trajectory, object tracking, overlap, object detection, aerial, pedestrian, occlusion, crowd