Image Inpainting

The ImageNet dataset contains 14,197,122 annotated images according to the WordNet hierarchy. Since 2010 the dataset is used in the ImageNet Large Scale Visual Recognition Challenge (ILSVRC), a benchmark in image classification and object detection. The publicly released dataset contains a set of manually annotated training images. A set of test images is also released, with the manual annotations withheld. ILSVRC annotations fall into one of two categories: (1) image-level annotation of a binary label for the presence or absence of an object class in the image, e.g., “there are cars in this image” but “there are no tigers,” and (2) object-level annotation of a tight bounding box and class label around an object instance in the image, e.g., “there is a screwdriver centered at position (20,25) with width of 50 pixels and height of 30 pixels”. The ImageNet project does not own the copyright of the images, therefore only thumbnails and URLs of images are provided. * Total number of non-empty WordNet synsets: 21841 * Total number of images: 14197122 * Number of images with bounding box annotations: 1,034,908 * Number of synsets with SIFT features: 1000 * Number of images with SIFT features: 1.2 million Source: ImageNet Large Scale Visual Recognition Challenge Image Source: https://cs.stanford.edu/people/karpathy/cnnembed/

CelebFaces Attributes dataset contains 202,599 face images of the size 178×218 from 10,177 celebrities, each annotated with 40 binary labels indicating facial attributes like hair color, gender and age. Source: Show, Attend and Translate: Unpaired Multi-Domain Image-to-Image Translation with Visual Attention Image Source: http://mmlab.ie.cuhk.edu.hk/projects/CelebA.html

Flickr-Faces-HQ (FFHQ) consists of 70,000 high-quality PNG images at 1024×1024 resolution and contains considerable variation in terms of age, ethnicity and image background. It also has good coverage of accessories such as eyeglasses, sunglasses, hats, etc. The images were crawled from Flickr, thus inheriting all the biases of that website, and automatically aligned and cropped using dlib. Only images under permissive licenses were collected. Various automatic filters were used to prune the set, and finally Amazon Mechanical Turk was used to remove the occasional statues, paintings, or photos of photos. Source: Flickr-Faces-HQ Dataset (FFHQ) Image Source: https://github.com/NVlabs/ffhq-dataset

The Places dataset is proposed for scene recognition and contains more than 2.5 million images covering more than 205 scene categories with more than 5,000 images per category. Source: Self-supervised Visual Feature Learning with Deep Neural Networks: A Survey Image Source: http://places.csail.mit.edu/browser.html

The CelebA-HQ dataset is a high-quality version of CelebA that consists of 30,000 images at 1024×1024 resolution. Source: IntroVAE: Introspective Variational Autoencoders for Photographic Image Synthesis Image Source: https://github.com/tkarras/progressive_growing_of_gans

VGGFace2 is a large-scale face recognition dataset. Images are downloaded from Google Image Search and have large variations in pose, age, illumination, ethnicity and profession. VGGFace2 contains images from identities spanning a wide range of different ethnicities, accents, professions and ages. All face images are captured "in the wild", with pose and emotion variations and different lighting and occlusion conditions. Face distribution for different identities is varied, from 87 to 843, with an average of 362 images for each subject.

The CASIA-WebFace dataset is used for face verification and face identification tasks. The dataset contains 494,414 face images of 10,575 real identities collected from the web. Source: On Hallucinating Context and Background Pixels from a Face Mask using Multi-scale GANs

ApolloScape is a large dataset consisting of over 140,000 video frames (73 street scene videos) from various locations in China under varying weather conditions. Pixel-wise semantic annotation of the recorded data is provided in 2D, with point-wise semantic annotation in 3D for 28 classes. In addition, the dataset contains lane marking annotations in 2D. Source: A2D2: Audi Autonomous Driving Dataset Image Source: https://arxiv.org/pdf/1803.06184.pdf

The Places365 dataset is a scene recognition dataset. It is composed of 10 million images comprising 434 scene classes. There are two versions of the dataset: Places365-Standard with 1.8 million train and 36000 validation images from K=365 scene classes, and Places365-Challenge-2016, in which the size of the training set is increased up to 6.2 million extra images, including 69 new scene classes (leading to a total of 8 million train images from 434 scene classes). Source: Semantic-Aware Scene Recognition Image Source: Places

Fashion-Gen consists of 293,008 high definition (1360 x 1360 pixels) fashion images paired with item descriptions provided by professional stylists. Each item is photographed from a variety of angles. Source: Fashion-Gen: The Generative Fashion Dataset and Challenge Image Source: Rostamzadeh et al

CASIA V2 is a dataset for forgery classification. It contains 4795 images, 1701 authentic and 3274 forged. Source: Copy-Move Forgery Classification via Unsupervised Domain Adaptation Image Source: https://www.mdpi.com/1099-4300/21/4/371

SEN12MS-CR is a multi-modal and mono-temporal data set for cloud removal. It contains observations covering 175 globally distributed Regions of Interest recorded in one of four seasons throughout the year of 2018. For each region, paired and co-registered synthetic aperture radar (SAR) Sentinel-1 measurements as well as cloudy and cloud-free optical multi-spectral Sentinel-2 observations from European Space Agency's Copernicus mission are provided. The Sentinel satellites provide public access data and are among the most prominent satellites in Earth observation. Source: Multi-Sensor Data Fusion for Cloud Removal in Global and All-Season Sentinel-2 Imagery

RICE is a remote sensing image dataset for cloud removal. The proposed dataset consists of two parts: RICE1 contains 500 pairs of images, each pair has images with cloud and cloudless size of 512512; RICE2 contains 450 sets of images, each set contains three 512512 size images, respectively, the reference picture without clouds, the picture of the cloud and the mask of its cloud. Source: https://github.com/BUPTLdy/RICE_DATASET

The Free-Form Video Inpainting dataset is a dataset used for training and evaluation video inpainting models. It consists of 1940 videos from the YouTube-VOS dataset and 12,600 videos from the YouTube-BoundingBoxes. Source: https://arxiv.org/abs/1904.10247 Image Source: https://github.com/amjltc295/Free-Form-Video-Inpainting

SEN12MS-CR-TS is a multi-modal and multi-temporal data set for cloud removal. It contains time-series of paired and co-registered Sentinel-1 and cloudy as well as cloud-free Sentinel-2 data from European Space Agency's Copernicus mission. Each time series contains 30 cloudy and clear observations regularly sampled throughout the year 2018. Our multi-temporal data set is readily pre-processed and backward-compatible with SEN12MS-CR.

A diverse dataset of human faces, including unconventional poses, occluded faces, and a vast variability in backgrounds. Source: DeepPrivacy: A Generative Adversarial Network for Face Anonymization

QST contains 1,167 video clips that are cut out from 216 time-lapse 4K videos collected from YouTube, which can be used for a variety of tasks, such as (high-resolution) video generation, (high-resolution) video prediction, (high-resolution) image generation, texture generation, image inpainting, image/video super-resolution, image/video colorization, image/video animating, etc. Each short clip contains multiple frames (from a minimum of 58 frames to a maximum of 1,200 frames, a total of 285,446 frames), and the resolution of each frame is more than 1,024 x 1,024. Specifically, QST consists of a training set (containing 1000 clips, totally 244,930 frames), a validation set (containing 100 clips, totally 23,200 frames), and a testing set (containing 67 clips, totally 17,316 frames). Click here (Key: qst1) to download the QST dataset.

PAL4Inpaint is a dataset consisting of 4,795 inpainting results with per-pixel perceptual artifacts annotations designed for image inpainting tasks.

The Inpainting dataset consists of synchronized Labeled image and LiDAR scanned point clouds. It's captured by HESAI Pandora All-in-One Sensing Kit. It is collected under various lighting conditions and traffic densities in Beijing, China.

Synthetic training set: This set is constructed in the following two steps and will be used for estimation/training purposes. i) 84,000 275 pixel x 400 pixel ground-truth fingerprint images without any noise or scratches, but with random transformations (at most five pixels translation and +/-10 degrees rotation) were generated by using the software Anguli: Synthetic Fingerprint Generator. ii) 84,000 275 pixel x 400 pixel degraded fingerprint images were generated by applying random artifacts (blur, brightness, contrast, elastic transformation, occlusion, scratch, resolution, rotation) and backgrounds to the ground-truth fingerprint images. In total, it contains 168,000 fingerprint images (84,000 fingerprints, and two impressions - one ground-truth and one degraded - per fingerprint). Synthetic test set: This set is constructed similarly to the synthetic training set and will be used to evaluate the reconstruction performance. In total, it contains 16,800 fingerprint images (8,400 fingerprints and two impressions - one ground-truth and one degraded - per fingerprint). Since this set will be used for the purpose of evaluating the reconstruction performance, only the degraded and not the ground-truth fingerprint images will be provided to participants. Real test set: This set is constructed by systematically drawing fingerprint images with varying sizes from publicly available datasets. In total, it contains 1680 fingerprint images (140 fingerprints and 12 impressions - high-quality scans under operational conditions - per fingerprint). Description from: Fingerprint inpainting and denoising (WCCI'18, ECCV'18)

During the covid-19 era wearing face masks posed new challenges to face-related tasks, including facial recognition, face inpainting, expression recognition, and object removal. Mask region segmentation is a preliminary stage to tackle the occlusion issue corresponding to the face-related tasks. Existing masked face datasets are not procedure binary segmentation maps because Segmenting mask regions manually is a time-consuming operation. As a result, existing unmasking methods; synthesize training data by overlaying masks on existing face datasets. However, since these techniques rely on an artificially generated mask, their effects tend to seem unnatural. To address this issue, the masked face segmentation dataset(MFSD) provides the first public training dataset for the mask segmentation task.

Inpainting networks are typically benchmarked on samples from Places2 dataset. However, this dataset does not have high resolution images for evaluation purposes. Instead, we will use images from the Unsplash-Lite Dataset, which contains 25k high resolution nature-themed photos. We randomly sampled 1000 images from the dataset. Each image is resized and cropped to 1024x1024, and a set of masks is generated with thin, medium, and thick brush strokes, using the methodology described in LaMa. The purpose of this dataset is to serve as a test-set for evaluating inpainting performance on high resolution natural images.

Dataset for the DREAMING - Diminished Reality for Emerging Applications in Medicine through Inpainting Challenge! The DREAMING Challenge is part of ISBI 2024. Participate for Awards, Money Prices, Proceedings Paper, Top-Tier Challenge Summary Paper! More information about the challenge and registration can be found on the challenge platform! The DREAMING training dataset contains 100 diverse, photorealistic surgery scenes with unique patients, surroundings and obstructions from dynamically moving and static surgical instruments and hands holding them. We also provide the masks defining the area to be inpainted for each image.

InpaintCOCO is a benchmark to understand fine-grained concepts in multimodal models (vision-language) similar to Winoground. To our knowledge InpaintCOCO is the first benchmark, which consists of image pairs with minimum differences, so that the visual representation can be analyzed in a more standardized setting. A data sample contains 2 images and 2 corresponding captions that differ only in one object, the color of an object, or the size of an object. The metric used in the paper compares if true image-text pairs are more similar than wrong image-text combinations and that for both image-text pairs: \begin{equation} \begin{split} \operatorname{sim}(i_\text{COCO},t_\text{COCO}) > \operatorname{sim}(i_\text{inp},t_\text{COCO}) \quad \land \ \operatorname{sim}(i_\text{inp},t_\text{inp}) > \operatorname{sim}(i_\text{COCO},t_\text{inp}) \end{split} \label{eq:challengeset} \end{equation} InpaintCOCO is published in Proceedings of the 3rd Workshop on Advances in Language and Vision Research (ALVR) at ACL 2024.