Papers with Code 아카이브

Image Classification

태스크의 하위 분류와 논문·코드·평가·기법·데이터셋을 탐색하세요.

하위 태스크
33
평가표
177
보존된 평가 기록
2,906

Image Classification is a fundamental task in vision recognition that aims to understand and categorize an image as a whole under a specific label. Unlike object detection, which involves classification and location of multiple objects within an image, image classification typically pertains to single-object images. When the classification becomes highly detailed or reaches instance-level, it is often referred to as image retrieval, which also involves finding similar images in a large database.

Source: Metamorphic Testing for Object Detection Systems

데이터셋

출처에 같은 태스크 이름으로 연결된 데이터셋입니다. 동명 태스크가 포함될 수 있으므로, 비교할 때는 평가 결과의 조건도 확인하세요.

LeafNet

The PlantVillage dataset, with over 54,000 images spanning 14 plant species and 26 disease types, has been widely used for leaf disease classification. However, it is limited in both scale and diversity. To address these limitations, we developed LeafNet, a large-scale dataset designed to support foundation models for leaf disease diagnosis. LeafNet comprises over 186,000 images from 22 crop species, covering 43 fungal diseases, 8 bacterial diseases, 2 mould (oomycete) diseases, 6 viral diseases, and 3 mite-induced diseases, categorized into 97 classes. The dataset was meticulously collected and processed to minimize intra-class variations while ensuring clarity by maintaining a consistent imaging distance. The disease symptom descriptions were curated from reputable sources, including UME, NIH, and published studies, providing high-quality annotations to support AI-driven plant pathology research.

LIB-HSI

The LIB-HSI dataset contains hyperspectral reflectance images and their corresponding RGB images of building façades in a light industrial environment. The dataset also contains pixel-level annotated images for each hyperspectral/RGB image. The LIB-HSI dataset was created to develop deep learning methods for segmenting building facade materials. The images were captured using a Specim IQ hyperspectral camera. The hyperspectral images are in the ENVI format. There are 393 training images, 45 validation images and 75 test images.

Lot-insts

LoT-insts contains over 25k classes whose frequencies are naturally long-tail distributed. Its test set from four different subsets: many-, medium-, and few-shot sets, as well as a zero-shot open set. To our best knowledge, this is the first natural language dataset that focuses on this long-tailed and open classification problem.

MineralImage5k

We present a comprehensive dataset comprising a vast collection of raw mineral samples for the purpose of mineral recognition. The dataset encompasses more than 5,000 distinct mineral species and incorporates subsets for zero-shot and few-shot learning. In addition to the samples themselves, some entries in the dataset are accompanied by supplementary natural language descriptions, size measurements, and segmentation masks. For detailed information on each sample, please refer to the minerals_full.csv file.

MM-Locate-News

MM-Locate-News is a dataset for location estimation of news. It consists of 6395 news articles covering 237 cities and 152 countries across all continents as well as multiple domains such as health, environment, and politics. The dataset is collected in a weakly-supervised manner, and multiple data cleaning steps are applied to remove articles with potential inaccurate geolocation information. The acquired dataset addresses drawbacks of other datasets such as BreakingNews as it considers multimodal content of news to label the corresponding location. Source: MM-Locate-News: Multimodal Focus Location Estimation in News Image Source: https://arxiv.org/pdf/2211.08042v1.pdf

No Background RGB Arabic Alphabets Sign Language Dataset

The AASL-Clear dataset is a collection of RGB images featuring Arabic alphabet sign Language gestures with backgrounds removed. Each image in this dataset showcases clear, isolated hand gestures, allowing for precise recognition and analysis of Arabic sign language alphabets. With transparent backgrounds, this dataset provides a clean and focused resource for training deep learning models in the domain of Arabic sign language recognition and classification. You can find the original dataset (RGB Arabic Alphabets Sign Language DatasetAASL) with the backgrounds Here: https://www.kaggle.com/datasets/muhammadalbrham/rgb-arabic-alphabets-sign-language-dataset

Notre-Dame Cathedral Fire

Number of images: 1,657 images during or after the fire If you use the dataset, please cite the following works: > Padilha, Rafael and Andaló, Fernanda A. and Rocha, Anderson. “Improving the chronological sorting of images through occlusion: A study on the Notre-Dame cathedral fire,” in 45th International Conference on Acoustics, Speech, and Signal Processing (ICASSP), 2020. ### Description of the event and data collection: On April 15th, 2019, large parts of Notre-Dame Cathedral's structure and spire were devastated by a fire. People worldwide followed the tragic event through images and videos that were shared by the media and citizens. From the generated imagery, we collected a total of 23,683 images posted on Twitter during and on the day after the fire. Even though most of them were related to the event, several were memes, cartoons, compositions and artwork, while some depicted the cathedral before the fire. As we focus on learning how the fire and appearance of the cathedral evolved during the event, we removed them, reducing our set to 5,206 relevant images. Among these, several examples were duplicates or near-duplicates of other images. Considering their little contribution to the training process, after their removal, we were left with 1,657 distinct images related to the event. The cleaning process involved using methods such as local sensitive hashing for filtering near-duplicates, and semi-supervised approaches based on Optimum-path Forest theory to mine for relevant and non-relevant imagery of the event. By analyzing the event's description, four main sub-events can be defined: spire on fire, spire collapsing, fire continues on roof, and fire extinguished. Each sub-event contains specific visual clues (e.g., the absence of the central spire) that can be leveraged to estimate the temporal position of an image. Each image in the dataset was manually labeled as being captured in one of these sub-events. We also consider an unknown category for images that do not contain any hint of the sub-event in which they were captured, such as zoom-ins of the cathedral's facades. Besides that, each image was annotated with respect to the intercardinal direction of the cathedral’s facade being depicted in the image (north, northeast, east, southeast, south, southwest, west, northwest). Image source: Improving the chronological sorting of images through occlusion: A study on the Notre-Dame cathedral fire

OLID I

The success of any AI-driven system relies heavily on vast amounts of training data. While AI applications in plant stress management have gained attention in recent years, there's still a significant lack of expert-annotated data, especially for tropical and subtropical crops. We're filling in this gap by releasing a public dataset with 4,749 leaf pictures of healthy, nutrient-deficient, and pest-affected tomatoes, eggplants, cucumbers, bitter gourds, snake gourds, ridge gourds, ash gourds, and bottle gourds. This dataset encompasses 57 unique classes, with high-resolution images (3024 x 3024) captured at three different sites in Bangladesh under natural field conditions. An expert panel from the Bangladesh Agricultural Research Institute (BARI) has labeled the images. This collection not only features the largest number of plant stress classes but also introduces the first multi-label classification challenge in the agricultural domain.

Omni-Image

Omni-Image is built as a challenging but tractable dataset for continual learning and few-shot learning. Omni-Image is a few-shot image classification task. It is a class-consistent subset of ImageNet images that mirrors the shallow-and-wide dataset shape of Omniglot. It contains 1000 classes with 20|100 images each, downsized to 84x84 pixels.

OpenStreetMap Multi-Sensor Scene Classification

A high-resolution multi-sensor remote sensing scene classification dataset, appropriate for training and evaluating image classification models in the remote sensing domain. The dataset consists of 8400 overhead scenes, each covered by Airbus Pléiades, Airbus SPOT, and USDA NAIP imagery. The scenes are classified into 12 OpenStreetMap categories: * man_made=bridge * man_made=breakwater * building=farm * power=substation * leisure=stadium * leisure=golf_course * waterway=dam * landuse=quarry * landuse=farmland * landuse=forest * natural=water * natural=bare_rock

Orchid2024

Orchid2024 is a fine-grained classification dataset specifically designed for Chinese Cymbidium orchid cultivars. It includes data collected from 20 cities across 12 provincial administrative regions in China and encompasses 1,269 cultivars from 8 Chinese Cymbidium orchid species and 6 additional categories, totaling 156,630 images. The dataset covers nearly all common Chinese Cymbidium cultivars currently found in China, with its fine granularity and focus on the real world making it a unique and practical resource for researchers and practitioners.

Photozilla

Photozilla is a large-scale dataset which includes over 990k images belonging to 10 different photographic styles. The dataset can be used to train classification models to automatically classify the images into the relevant style.

PoTATO: A Dataset for Analyzing Polarimetric Traces of Afloat Trash Objects

PoTATO is a dataset designed to enhance the detection of floating plastic waste in aquatic environments by leveraging polarimetric imaging. It comprises 12,380 labeled images of plastic bottles, each enriched with detailed polarimetric information. The dataset aims to improve object detection accuracy under challenging outdoor lighting conditions and water surface reflections. By providing raw image data, PoTATO offers researchers the opportunity to explore novel approaches and advance state-of-the-art object detection algorithms. The dataset and associated code are publicly available

PRImA

The Prima head pose dataset consists of 2790 images of 15 persons recorded twice. Pitch values lie in the interval [−60∘,60∘], and yaw values lie in the interval [−90∘,90∘] with a 15∘ step. Thus, there are 93 poses available for each person. All the recordings were achieved with the same background. One interesting feature of this dataset is the pose space is uniformly sampled. The dataset is annotated such that a face bounding box (manually annotated) and the corresponding yaw and pitch angle values are provided for each sample. Source: Robust Head-Pose Estimation Based on Partially-Latent Mixture of Linear Regressions Image Source: http://www-prima.inrialpes.fr/perso/Gourier/Faces/HPDatabase.html

PTVD

PTVD is a plot-oriented multimodal dataset in the TV domain. It is also the first non-English dataset of its kind. Additionally, PTVD contains more than 26 million bullet screen comments (BSCs), powering large-scale pre-training.

RawRipe Dataset

The dataset comprises of images of 10 types of fruits in both raw and ripe states. The fruit-types include: apple, banana, coconut, gauva, litchi, mango, orange, papaya, pomegranate and strawberry. The dataset can be used two solve three types classification problem: 1) raw vs ripe for a specific fruit (agricultural perspective) 2) raw vs ripe for any fruit (market perspective) 3) multi-class/label classification (automation perspective) For more details, refer to the following paper: Fruit Maturity Recognition from Agricultural, Market and Automation Perspectives, IECON'21 Paper link: https://ieeexplore.ieee.org/document/9589215 Please cite the paper if you use the dataset in your projects/papers. @INPROCEEDINGS{9589215, author={Rao Jerripothula, Koteswar and Kumar Shukla, Sarvesh and Jain, Samyak and Singh, Shudhanshu}, booktitle={IECON 2021 – 47th Annual Conference of the IEEE Industrial Electronics Society}, title={Fruit Maturity Recognition from Agricultural, Market and Automation Perspectives}, year={2021}, pages={1-6}, doi={10.1109/IECON48115.2021.9589215} }

RMOT-223

In this dataset, various objects are arranged on a white table. A UR5e robot picks and place a target object specified on the title of the video/image sequence. Videos under auto- folder are collected with automatic operation of the robot. Videos under human- folders are collected with the tele-operation of the robot. Ground-truth tracking bounding boxes are generated with STARK, and when the target exits the camera frame, the bounding box estimation is switched to [-1, -1, -1, -1], indicating target not shown.

ShipSpotting

To construct such a dataset, a straightforward approach was scraping images from the web. The main source for our dataset is ShipSpotting, which serves as a repository for user uploaded images, hosting a vast collection of ship images, amounting to approximately 3 million. Furthermore, for each image, valuable supplementary information is available, such as the type of the ship, and present and past names. Next, we made sure that as many images as possible were collected in our dataset, since in deep learning, the quantity of training data directly influences the quality of results. A larger volume of data enables models to generalize more effectively. Thus we scrape all the images and as a result, the dataset comprises a total of 1.517.702 samples. We exclude many classes of ships from our final analysis and concentrate on the more common and valuable for a real scenario use case. The total number of different classes is 20 and the ship categories included are Bulkers, Containerships, Cruise ships, Dredgers, Fire Fighting Vessels, Floating Sheerlegs, General Cargo, Inland, Livestock Carriers, Passenger Vessels, Patrol Forces, Reefers, Ro-ro, Supply ships, Tankers, Training ships, Tugs, Vehicle Carriers, Wood Chip Carriers. The total amount of samples after this class selection is 507.918.

SolarDK

SolarDK is a dataset for the detection and localization of solar. It comprises images from GeoDanmark with a variable Ground Sample Distance (GSD) between 10 cm and 15 cm, all sampled between March 1st and May 1st during 2021, containing 23,417 hand labelled images for classification and 880 segmentation masks, in addition to a set of about 100,000+ images for classification covering most variations of Danish urban and rural landscapes. Source: SolarDK: A high-resolution urban solar panel image classification and localization dataset Image Source: https://arxiv.org/pdf/2212.01260v1.pdf

SPOT-10

The SPOTS-10 dataset is an extensive collection of grayscale images showcasing diverse patterns found in ten animal species. Specifically, SPOTS-10 contains 50,000 32 x 32 grayscale images, divided into ten categories, with 5,000 images per category. The training set comprises 40,000 images, while the test set contains 10,000 images. The SPOTS-10 dataset is freely available on the project GitHub page: https://github.com/amotica/spots-10 by cloning the repository.

SSBI Dataset

The Synthetic Signature Bankcheck Images (SSBI) Dataset is the first publicly available dataset of bank check images with annotations for detecting handwritten components, including names, amounts, dates, and signatures. It also supports both writer-independent and writer-dependent signature verification tasks by providing labels for genuine and forged signatures and IDs of the signature authors. This dataset contains 2180 synthetic bank check images with: -1680 genuine signatures - 500 forged signatures The signatures are from 19 different signers. In addition to the signature bounding boxes, the dataset also contains the bounding boxes for the date, amount, and payee fields.

SuSy Dataset

The SuSy Dataset combines authentic photographs and AI-generated images designed for training and evaluating synthetic image detection models. It contains over 25,000 images from six different sources, including real-world photographs from COCO and synthetic images created by state-of-the-art diffusion models such as DALL-E 3, Midjourney, and Stable Diffusion. ## Authentic Images - COCO (Common Objects in Context): A large-scale object detection, segmentation, and captioning dataset. It includes over 330,000 images, with 200,000 labeled using 80 object categories. For this dataset, we use a random subset of 5,435 images. License: Creative Commons Attribution 4.0 license ## Synthetic Images - dalle-3-images: Contains 3,310 unique images generated using DALL-E 3. The dataset does not include the prompts used to generate the images. License: MIT license - diffusiondb: A large-scale text-to-image prompt dataset containing 14 million images generated by Stable Diffusion 1.x series models (2022). We use a random subset of 5,435 images. License:** CC0 1.0 Universal license - realisticSDXL: Contains images generated using the Stable Diffusion XL (SDXL) model released in July 2023. We use only the "realistic" category, which contains 5,435 images. License: CreativeML OpenRAIL-M license - midjourney-tti: Contains images generated using Midjourney V1 or V2 models (early 2022). The original dataset provided URLs, which were scraped to obtain the images. License: CC0 1.0 Universal license (for links only, images are property of users who generated them) - midjourney-images: Contains 4,308 unique images generated using Midjourney V5 and V6 models (2023). License: MIT license

SUT

This paper introduces a new large-scale dataset for Farsi document images, named SUT, which aims to tackle the challenges associated with obtaining diverse and substantial ground-truth data for supervised models in document image analysis (DIA) tasks, like document image classification, text detection and recognition, and information retrieval. The dataset comprises 62,453 images that have been categorized into 21 distinct classes, including identity documents featuring synthetically generated personal information superimposed on various backgrounds. The dataset also includes corresponding files with labeling information for the images. The ground-truth data is organized in CSV files containing image file paths and associated information about the embedded data.

SVLD

The social vision and language dataset is a large-scale multimodal dataset designed for research into social contextual learning. Source: A Dataset and Benchmarks for Multimedia Social Analysis Image Source: https://cannylab.github.io/svld/

Synthetic COVID-19 CXR Dataset

A public open dataset of synthetic chest X-ray images of COVID-19. The dataset consists of 21,295 synthetic COVID-19 chest X-ray images. Images are generated using an unsupervised domain adaptation approach by leveraging class conditioning and adversarial training from source datasets RSNA Kaggle Dataset and COVID-19 Image Data Collection. Implementation of the algorithm is available here.

TCB-DS

The TCB-DS dataset is a specialized collection of microscopic images focusing on the automatic recognition of cyanobacteria genera. This dataset was meticulously compiled to address the challenges associated with the varying image qualities due to differences in contrast, resolution, size, lighting, and the presence of noise in the original images. It includes 2,591 images with varying dimensions, ranging from a minimum of 11 × 41 pixels to a maximum of 5184 × 3456 pixels. After using this dataset please cite: Automated identification of toxigenic cyanobacterial genera for water quality control purposes https://doi.org/10.1016/j.jenvman.2024.121274

Tecnalia WEEE HYPERSPECTRAL DATASET

Tecnalia Hyperspectral Dataset contains different non-ferreous fractions of Waste from Electric and Electronic Equipment (WEEE) of Copper, Brass, Aluminum, Stainless Steel and White Copper. Images were captured by a hyperspectral Specim PHF Fast10 camera that is able to capture wavelengths in the range 400 to 1000 nm with a spectral resolution of less than 1 nm. The PHF Fast10 camera is equipped with a CMOS sensor (1024 × 1024 resolution), a Camera Link interface and a special Fore objective OL10. The provided dataset contains 76 uniformly distributed wave-lengths in the spectral range [415.05 nm, 1008.10 nm]. Illumination setup, as described in \cite{picon2012real}, was specifically designed to reduce the specular reflections generated by the surface of the non-ferrous materials and to provide a homogeneous and even illumination that covers the wavelengths sensitive to the hyperspectral camera. The illumination system consists of a parabolic surface that uniformly distributes the light generated by 9 halogens and 18 white LEDs covering the spectral range between 400 to 1000 nm.

TEM nanowire morphologies for classification and segmentation

TEM image dataset containing four nanowire morphologies of bio-derived protein nanowires and synthetic peptide nanowires. The peptide / protein nanowires used in this study were synthesized and imaged by Brian Montz in Prof. Todd Emrick's research group at the Department of Polymer Science and Engineering Department, University of Massachusetts Amherst. We acknowledge financial support from the U.S. National Science Foundation, Grant NSF DMREF #1921839 and DMREF #1921871. Nanowires were classified into either of the four morphologies: bundle, singular, dispersed or network. Each morphology contains 100 images (jpg files). For the dispersed and network morphologies, because these two morphologies are harder to visually distinguish, we have created manual segmentation labels of the nanowires (included in these two morphology folders as png files). Percolation analysis was done on these manually segmented nanowires to provide quantitative metric on whether the nanowires form a network in the image. encoders_trained_with_optimized_hyperparameter.zip contains 4 sets of encoders trained with either SimCLR or Barlow-Twins self-supervised methods on either generic TEM images, or generic everyday photographic images (each with 5 replicates with different random seed) with optimized hyperparameters. seg_mask_5_resolutions.zip contains ground truth 2D binary encoding of segmented nanowires at 5 resolutions.

TLF2K

Table-LastFm2K (TLF2K) is a relational table dataset derived from the classical LastFM2K dataset. It contains three tables: artists, user_artists, and user_friends. Notably, the artists table has been enhanced with more detailed features, and the tags for each artist have been streamlined. The dataset also provides a standard classification task for music genre classification of artists.

topex-printer

topex-printer is a dataset containing 102 machine parts of a label printing machine. It includes these parts for two domains, real photos and CAD rendered models.

출처: Papers with Code · CC BY-SA 4.0. 아카이브 자료와 새로 수집한 논문은 별도로 관리합니다.