High-quality Computer Vision Dataset to Train Your AI Model

Description: Internet collected animal images in variable scenarios like indoor, outdoor, nature, gardon and so on.

Description: Arabic & Thai & Vietnamese & Hindi & English & Chinese Language Dataset

Description: The Arabic Text Dataset contains a collection of text samples written in Arabic. It includes various forms of content, such as news articles, social media posts, literature, and dialogue, spanning different topics and writing styles. This dataset is used for tasks such as natural language processing (NLP), text classification, sentiment analysis, and machine translation in Arabic language applications.

Description: The "Asian Face Occlusion Dataset" is tailored for the visual entertainment industry, comprising a vast collection of internet-collected images, each with a resolution exceeding 2736 x 3648 pixels. This dataset focuses on instance and semantic segmentation of Asian faces, specifically targeting individuals aged between 18 and 50 with a male-to-female ratio of 3:7. The unique aspect of this dataset is the inclusion of various face-covering items, providing a diverse range of occlusion scenarios.

Description: The "Asian Single ID Photo Matting Dataset" is curated for the visual entertainment and social networking service (SNS) sectors, featuring a collection of internet-collected Asian face ID photos, all with a high resolution of 6720 x 4480 pixels. This dataset focuses on contour segmentation, offering pixel-level segmentation specifically tailored to Asian facial features in ID photos, facilitating precise face recognition and editing applications.

Description: The "Asian Student Classroom Emotions Dataset" is specifically designed for educational applications, featuring internet-collected images of Asian students in classroom settings, all at a uniform resolution of 1280 x 720 pixels. This dataset employs bounding box annotations and classification techniques to identify and categorize students' emotional and performance states in the classroom, aiming to enhance educational methodologies and student engagement strategies.

Description: The "Asian Style Gestures Dataset" is curated for the visual entertainment industry, featuring a collection of internet-collected images with resolutions ranging from 530 x 360 to 2973 x 3968 pixels. This dataset specializes in annotations of hands displaying Asian style gestures, such as nods, hearts, rock, OK, putting hands together, clasping hands, etc., utilizing bounding boxes and tags for precise identification.

Description: The Bank Cheque Dataset (Document AI): Synthetic bank cheques consists of artificially generated cheque images designed to replicate the appearance and content of real cheques. It includes various elements such as payee names, amounts, dates, signatures, and cheque numbers. This dataset is used for training and evaluating Document AI systems in tasks like optical character recognition (OCR), cheque processing, and automated data extraction, providing a controlled environment for model development without the privacy concerns of real cheques.

Recording Condition: - Clicked Images - Scanned - Web scrapper

Description: The Bank Statement Dataset (Document AI): Synthetic bank statements includes artificially generated bank statements designed to simulate real financial documents. It features various transaction records, dates, amounts, and account details, structured to mirror real-world formats and content. This dataset is used for training and evaluating Document AI systems in tasks such as optical character recognition (OCR), data extraction, and document analysis, offering a controlled environment without the privacy issues of actual financial data.

Recording Condition: - Scanned - Bank_Statement - Web scrapper

Description: Barcode Tye: Code128, UPC/EAN, DataMatrix, PDF417, Aztec, Multi-code

Recording Device: Honor 9A, Huawei mate 10 pro, iPad, iPhone (6S, 7 Plus, SE, X, 11, 12, 12 mini, 12 Pro Max), Moto (E4, onepower), One plus (6T, 7T, One), Oppo A3s, Real Me, Samsung (A20, A30, A32, M12, M31), Vivo z1pro, Xiaomi Mi10T+

Recording Condition: - Bright_Indoor - Low_Indoor - Low_Outdoor - Normal - Sunny

Description: The "Blur Area Segmentation Dataset" is designed for use in robotics and visual entertainment, composed of internet-collected images with resolutions ranging from 960 x 720 to 1024 x 768 pixels. This dataset focuses on semantic segmentation, specifically targeting blue areas within images. Each blue area is annotated at the pixel level, providing valuable data for applications requiring color-based segmentation or analysis.

Description: The "Car Key Point Identification Dataset" is designed for applications in visual entertainment and autonomous driving, featuring a collection of internet-collected images with a resolution of 640 x 512 pixels. This dataset employs bounding boxes to identify target cars and annotates 14 key points on each vehicle, including the four top points, the four lights, the four wheels, and the glass areas on the front and left side, providing detailed data for car modeling and recognition tasks.

Description: The "Cat & Dog Segmentation Dataset" is crafted for the media & entertainment and tourism industries, featuring a broad collection of internet-collected images with resolutions varying from 367 x 288 to 3456 x 4608 pixels. This dataset focuses on contour segmentation and includes diverse annotations such as humans, cats, dogs, and environmental elements like walls, tables, grass, and water surfaces, among others.

Description: The "Cat & Dog Body Segmentation Supplementary Dataset" is tailored for the visual entertainment industry, comprising a variety of internet-collected images with resolutions exceeding 440 x 440 pixels. This dataset focuses on contour segmentation, specifically delineating the outlines of cats and dogs of various breeds, providing detailed data for applications requiring precise pet representations.

Description: The "CCTV Traffic Scene Semantic Segmentation Dataset" offers a unique perspective for autonomous driving development, capturing the intricacies of traffic scenes from a stationary point of view. Utilizing high-resolution CCTV footage from road monitoring cameras, with resolutions exceeding 1600 x 1200 pixels and a frame rate of over 7 fps, this dataset provides detailed instance segmentation of various elements in traffic, including humans, animals, cycling vehicles, automobiles, and road barriers. It also encompasses a range of weather conditions, offering a robust dataset for training AI systems to understand and interpret diverse traffic scenarios from a fixed vantage point.

Description: The "Characters Contour Segmentation Dataset" is specifically designed for Optical Character Recognition (OCR) applications, featuring a collection of internet-collected images with resolutions ranging from 461 x 169 to 1080 x 1350 pixels. This dataset is centered around contour segmentation, focusing on the precise delineation of OCR optical characters to facilitate accurate character recognition and text extraction processes.

Description: The "Characters Relationship Segmentation Dataset" is designed for the robotics and visual entertainment industries, featuring a wide range of internet-collected images with resolutions spanning from 1280 × 720 to 4608 × 3456. This unique dataset focuses on the relationships between humans, and between humans and objects, providing valuable insights for interaction dynamics.

Description: Chinese & English & Tibetan & Uyghur Language Dataset

Description: The Chinese and English Menu Dataset contains images or text samples of restaurant menus that feature both Chinese and English languages. It includes various fonts, layouts, and menu structures, presenting bilingual dish names, descriptions, and prices. This dataset is useful for tasks such as optical character recognition (OCR), machine translation, and menu digitization in multilingual settings.

Description: The Chinese Bills Dataset includes images or text samples of various types of bills, such as invoices, receipts, and statements, written in Chinese. It features diverse formats and content, including item descriptions, amounts, and dates. This dataset is used for tasks like optical character recognition (OCR), financial document processing, and automated data extraction.

Description: The Chinese Handwritten Composition Dataset contains samples of handwritten Chinese text, including compositions, essays, and other long-form text. It features various handwriting styles and levels of complexity, and is used for tasks such as handwriting recognition, text analysis, and machine learning model training.

Description: The Chinese WIFI Prompt Dataset consists of text samples found in WIFI prompts and login screens written in Chinese. It typically includes various prompts, instructions, and error messages related to connecting to or managing WIFI networks. This dataset is used for tasks like text recognition, natural language processing, and improving user interfaces for network connectivity.

Description: The "City Sky Contour Segmentation Dataset" is curated for the visual entertainment sector, featuring a collection of internet-collected images with a high resolution of 3000 x 4000 pixels. This dataset is dedicated to contour segmentation, focusing on capturing the sky in urban settings with elements such as buildings and plants, providing a detailed backdrop for various visual content creation.

Description: The "Clothes Segmentation Dataset" is crafted for the e-commerce, fashion, and visual entertainment sectors, incorporating a wide array of internet-collected images with resolutions ranging from 183 x 275 to 3024 x 4032 pixels. This dataset specializes in contour and semantic segmentation, featuring around 30 target categories including clothing items, accessories, and body parts, facilitating detailed analysis and application in fashion technology.

Description: The "Clothing Classification Dataset" is an essential resource for the fashion, e-commerce, and digital marketing industries, aiming to streamline the online shopping experience. This dataset encompasses a wide array of clothing items collected from the internet, covering various scenarios such as e-commerce websites, fashion shows, social media platforms, and offline user-generated content. It's designed to support the development of sophisticated algorithms for clothing classification, trend analysis, and personalized recommendation systems.

Description: The "Clothing Keypoints Dataset" aims to enhance the precision of fashion-related AI applications by providing a large-scale collection of images for keypoint detection tasks. This dataset includes internet-collected images that span a wide array of scenarios, including e-commerce platforms, fashion shows, social media, and offline user-generated content. It is meticulously annotated to identify keypoints on clothing items, facilitating the development of algorithms for pose estimation, size fitting, style matching, and interactive shopping experiences. The dataset includes classified labels, bounding boxes, and keypoints for 80 different clothing types, making it a comprehensive resource for improving the accuracy and reliability of fashion AI systems.

Description: The "Clothing Pattern Classification Dataset" is specifically designed to address the needs of the fashion industry, focusing on the classification of various clothing patterns. This dataset gathers internet-collected images that showcase clothing from different scenarios such as e-commerce platforms, fashion shows, social media, and offline user-generated content. It aims to facilitate the development of AI models that can accurately recognize and classify over 30 common clothing patterns, enhancing online shopping experiences and supporting trend analysis.

Description: The "Clothing Segmentation and Fabrics Classification Dataset" merges the complexity of clothing segmentation with the specificity of fabric classification, offering a dual-purpose dataset for the fashion industry. It includes internet-collected images from a variety of sources such as e-commerce websites, fashion shows, social media, and offline user-generated content. The dataset is structured to support the development of AI models that can perform both detailed segmentation of clothing items and classify them into 11 common fabric categories, encompassing 80 distinct clothing types. This dual approach aims to enhance online shopping experiences by providing detailed insights into the type of clothing and fabric, facilitating better inventory management and personalized shopping recommendations.

Description: The "Clothing Segmentation Dataset" is designed to propel the capabilities of AI in the fashion industry by providing a comprehensive collection of images for semantic segmentation tasks. This dataset encompasses internet-collected images from various scenarios such as e-commerce platforms, fashion shows, social media, and offline user-generated content. It focuses on enabling precise segmentation of clothing items, including main human parts, clothing pieces, and accessories, to support the development of advanced AI models for automated image analysis and product categorization.

Description: The "Cloudy Day City Road Dash Cam Video Dataset" is crafted to address the challenges autonomous driving systems face in overcast weather conditions. Captured with driving recorders at a resolution exceeding 1920 x 1080 pixels and a frame rate of over 31 fps, this dataset ensures detailed visibility even under the diffused lighting of cloudy skies. It includes bounding boxes and tags for more than 10 object categories commonly encountered in urban settings, such as humans, cars, electric bicycles, vans, and trucks. This dataset aims to refine AI models' ability to navigate and make informed decisions in less-than-ideal weather conditions, enhancing safety and reliability.

Description: The "Cloudy Day Crossroad Dash Cam Video Dataset" specifically captures the intricate dynamics of crossroad navigation under cloudy weather conditions. This dataset is filmed with high-resolution driving recorders, boasting resolutions over 1920 x 1080 pixels and a frame rate of more than 32 fps, to ensure clarity and detail even in subdued lighting. It annotates more than 10 typical urban object categories, including humans, cars, electric bicycles, vans, and trucks, amidst the unique challenges presented at crossroads during cloudy days. The dataset is an essential resource for developing autonomous driving systems capable of understanding and reacting appropriately to complex urban intersections, especially when visibility is affected by overcast skies.

Description: The "Common Objects Segmentation Dataset" serves the e-commerce and visual entertainment industries with a broad collection of internet-collected images, featuring resolutions ranging from 800 × 600 to 4160 × 3120. This dataset covers a wide array of everyday scenes and objects, including people, animals, furniture, and more, annotated for both instance and semantic segmentation.

Description: The "Damaged Board Parts Segmentation Dataset" is a niche collection tailored for the manufacturing sector, especially in wood and board production. It features internet-collected images with high resolutions ranging from 3024 x 4032 to 2048 x 5750 pixels. This dataset focuses on semantic segmentation of various types of board damage, including cracks, insect damage, and decay, aiding in quality control and manufacturing processes.

Description: 360 degrees walk around videos of cars with damages at a normal, steady pace with top and bottom always visible Damage: a scratch, dent, ding, or crack that is larger than a golf ball in length Outer Panel Damage: bumpers, fenders, quarter panels, doors, hoods, and trunks Location: Asia, US, Canada, and Europe

Recording Device: Mobile Camera

Recording Condition: Mixed Lighting Conditions

Description: 490+ cars and 3958 car photos with annotated images (along with metadata) of damaged cars. Covers all sides of the car (8 photos for each car) - Insurance Claim Process Use Cases.

Recording Device: Mobile Camera

Recording Condition: Mixed Lighting Conditions

Description: The "Dashcam Traffic Scenes Semantic Segmentation Dataset" is essential for pushing the boundaries of autonomous driving technologies. This dataset contains driving recorder images with a resolution of about 1280 x 720 pixels, segmented semantically to reflect various elements of urban and suburban traffic environments. It comprehensively categorizes 24 different objects and scenarios, including sky, people, motor vehicles, non-motorized vehicles, highways, pedestrian paths, zebra crossings, trees, buildings, and more. This detailed semantic segmentation allows autonomous driving systems to better understand and interpret the complexities of the road, enhancing navigation and safety protocols.

Description: The "Drivable Area Segmentation Dataset" is meticulously crafted to enhance the capabilities of AI in navigating autonomous vehicles through diverse driving environments. It features a wide array of high-resolution images, with resolutions ranging from 1600 x 1200 to 2592 x 1944 pixels, capturing various pavement types such as bitumen, concrete, gravel, earth, snow, and ice. This dataset is vital for training AI models to differentiate between drivable and non-drivable areas, a fundamental aspect of autonomous driving. By providing detailed semantic and binary segmentation, it aims to improve the safety and efficiency of autonomous vehicles, ensuring they can adapt to different road conditions and environments encountered in real-world scenarios.

Description: The "E-commerce Product Dataset" is a comprehensive collection tailored for the e-commerce sector, featuring a wide range of products from 16 main categories including shoes, hats, bags, furniture, digital products, jewelry, and more. With over 200k SKUs, this dataset is equipped with bounding boxes and category tags, making it a pivotal resource for product classification and inventory management.

Description: Our "Eastern Asia Single-person Portrait Matting Dataset" targets the nuanced requirements of the fashion, internet, and entertainment sectors, featuring single-person portraits from Eastern Asia in a variety of settings including indoor, outdoor, street, and sport. This dataset is specially curated for pixel-level fine segmentation tasks, capturing diverse postures and scenarios.

Description: The English Menu Dataset includes images or text samples of restaurant menus written in English. It features a variety of fonts, layouts, and formatting styles, with content ranging from dish names to descriptions and prices. This dataset is often used for tasks like optical character recognition (OCR), text extraction, and menu digitization in food-related applications.

Description: The English Scenes Text Dataset consists of images containing natural scenes with embedded English text. The text appears in various forms, such as signs, billboards, and posters, often in diverse fonts, sizes, and orientations. This dataset is commonly used for training and testing models in text detection, recognition, and scene understanding tasks.

Description: The English & Chinese Handwriting Dataset contains handwritten samples in both English and Chinese, showcasing various writing styles and character complexities. It is typically used for training and evaluating handwriting recognition models, supporting multilingual text analysis, and other related research. The dataset includes a diverse range of characters, digits, words, and sentences in both languages.

Description: The English & Chinese Shopsign Dataset includes images of shop signs that feature both English and Chinese text. It captures various signage elements such as store names, advertisements, promotions, and directions, displayed in diverse fonts, styles, and formats. This dataset is used for tasks like text detection and recognition, multilingual scene understanding, and improving computer vision models for interpreting bilingual signage.

Description: The English & Chinese Special Angle Text Dataset contains images of text displayed at various angles and orientations in both English and Chinese. It includes text from sources like signs, advertisements, and documents that are not presented in standard horizontal formats. This dataset is used for training and evaluating text detection and recognition models, particularly those capable of handling text in non-traditional orientations and perspectives.

Description: The "Escalator Face Bounding Dataset" is specifically designed for use in government and security sectors, featuring a collection of outdoor-collected images with resolutions exceeding 960 x 540 pixels. This dataset employs bounding boxes to annotate the head, face, and entire body of individuals captured in escalator settings. The annotations are meticulously drawn to encompass the entire face, including any masks that might be worn, ensuring comprehensive facial recognition capabilities even in partially obscured conditions.

Description: The "Human Body Semantic Segmentation Dataset" serves the fashion, internet, and entertainment sectors with a diverse collection of human body images. This dataset, featuring an even distribution across genders and ages from various countries, is ideal for applications requiring detailed analysis of human postures, hairstyles, and different scenarios. With fine labeling of 19 human body areas, it facilitates advanced semantic segmentation tasks.

Description: The "Facial 17 Parts Segmentation Dataset" is specifically compiled for the visual entertainment industry, featuring a range of internet-collected facial images with resolutions exceeding 1024 x 682 pixels. This dataset is dedicated to semantic segmentation, delineating 17 facial categories such as eyebrows, lips, eye pupils, and more. It also includes a selection of portrait images with occlusions, adding complexity and diversity to the dataset for more realistic application scenarios.

Description: The "Facial Color Segmentation Dataset" is tailored for the beauty and visual entertainment sectors, consisting of internet-collected images with resolutions from 1028 x 1028 to 6016 x 4016 pixels. This dataset focuses on semantic segmentation of facial skin colors, including black, yellow, white, and brown, facilitating diverse applications in cosmetics, virtual makeovers, and inclusive digital content.

Description: The "Facial Parts Semantic Segmentation Dataset" supports the beauty and media & entertainment sectors, with a collection of images sourced both online and offline. Resolutions vary from 300 x 300 to 4480 x 6720, covering comprehensive facial area categories such as eyes, eyebrows, nose, mouth, hair, and accessories, each meticulously annotated for semantic segmentation and bounding box tasks.

Description: Facial recognition datasets consist solely of images of faces, with no additional annotations. They include diverse examples of facial features, poses, and lighting conditions, and are used to train and evaluate facial recognition systems for tasks like face detection and recognition.

Recording Condition: Lighting Condition: - Bright Light Or Sunlight - Shade Or Overcast - Night Or Dim Light

Description: The "Flying Wire Segmentation Dataset" is specifically developed for the visual entertainment industry, comprising internet-collected images with resolutions exceeding 1024 x 638 pixels. This dataset is focused on instance segmentation, with a primary emphasis on annotating ropes or wires that span between buildings, offering valuable data for creating realistic urban environments in digital content.

Description: Our "Food Contour Matting Dataset" enriches the culinary and visual content domains, featuring ~200 food types from global cuisines. It's designed for businesses in catering, tourism, and entertainment, offering personalized experiences through detailed segmentation annotations.

Description: The "Food Segmentation Dataset" serves the tourism and visual entertainment sectors, consisting of a curated selection of internet-collected images with resolutions from 256 x 256 to 1024 x 768 pixels. This dataset is dedicated to contour segmentation, focusing on common foods and their accompanying plates or bowls, facilitating detailed analysis and representation in various applications.

Description: The "Full Body Clothing Classification Dataset" is specifically curated to support the advancement of AI in recognizing and classifying full-body clothing from a wide range of internet-collected images. With a focus on high-resolution images, specifically 768 x 1024 pixels, this dataset aims to enhance the precision in classifying full-body attire into major categories such as tops, pants, and skirts, further delineating into 30 sub-categories including jackets, sportswear, baseball uniforms, sweaters, sweatpants, jeans, and half skirts, among others. This dataset is designed to facilitate the development of sophisticated AI models that can accurately classify complex clothing types in full-body images, thereby improving the efficiency and user experience of online fashion retail.

Description: Sets of still images taken in either daytime or nighttime settings where natural or artificial lighting create a digital artifact known as a ghost.

Recording Device: iPhone & iPad Camera

Recording Condition: - Day Time - Night Time

Description: The "Glasses Segmentation Dataset" is aimed at the apparel and visual entertainment sectors, incorporating a diverse array of internet-collected images with resolutions from 165 x 126 to 1250 x 1458 pixels. This dataset focuses on semantic segmentation of various types of eyewear, including pure transparent glasses, sunglasses, and translucent glasses, providing detailed annotations for each category.

Description: The "Hair Semantic Segmentation Dataset" serves the apparel and media & entertainment industries, featuring a curated collection of internet-collected images with resolutions varying from 343 x 358 to 2316 x 3088 pixels. This dataset specializes in high-precision contour and semantic segmentation of hair, offering detailed annotations for a wide range of hairstyles and textures.

Description: The "Hand Key Point Skeleton Dataset" is designed for applications in visual entertainment and augmented/virtual reality (AR/VR), featuring a collection of indoor-collected images with a high resolution of 3024 x 4032 pixels. This dataset focuses on labeling 21 key points of the hand skeleton, capturing specific single-handed or two-handed poses such as forming a heart shape, placing a hand on the cheek, stretching, and more.

Description: Live Photos with Handwritten text for Japanese, Korean & Russian

Recording Device: iPhone & iPad Camera

Recording Condition: - Aggressive Lighting/Glare - Camera Flash On - Colored Light - Low Light, No Camera Flash - Normal

Description: The "Head and Neck Semantic Segmentation Dataset" is designed for the e-commerce & retail and media & entertainment sectors, featuring a collection of AI-generated cartoon images with resolutions above 1024 x 1024 pixels. This dataset focuses on semantic segmentation, specifically targeting the main character's head, including face, hair, and any accessories, as well as the neck area up to the collarbone, with an allowance for small, unsegmented parts on the edges.

Description: Collect images (1 Enrollment photo, 20 Historical photos per Identity) and videos (1 Indoor, 1 Outdoor) from unique identities

Description: The "Human And Accessories Segmentation Dataset" is a valuable resource for the apparel, e-commerce, and media & entertainment industries, featuring internet-collected images with resolutions ranging from 584 x 429 to 3744 x 5616. This dataset is rich in diversity, encompassing a wide array of accessories like mobile phones, suitcases, skateboards, and animals, all annotated for semantic segmentation.

Description: The "Human And Multi-object Panoptic Segmentation Dataset" is curated for applications in visual entertainment, featuring a wide array of internet-collected images with resolutions exceeding 1280 x 700 pixels. This comprehensive dataset integrates both instance and semantic segmentation to label a diverse range of elements found in everyday life, including natural scenery, people, buildings, and animals, offering a panoptic view of various scenes and subjects.

Description: The "Human Body High Precision Segmentation Dataset" is a comprehensive collection aimed at the apparel, e-commerce, and visual entertainment sectors, combining manually shot and internet-collected images with resolutions from 316 × 600 to 6601 × 9900. It focuses on high-precision segmentation of the human body, capturing intricate details of limbs, clothing, facial features, skin, and accessories.

Description: Images are from internet. Resolution ranges from 105 x 251 to 319 x 951.