Loading...
Loading...
Image classification, object detection, segmentation, face recognition, OCR, image generation, video understanding
16,460 datasets
Kaggle hosts a dataset of over 83,600 captcha images with labels for optical character recognition (OCR) model training. The images contain 3-character alphanumeric sequences. The author, organization, and collection method are not specified.
Supporting 6+ annotation formats including COCO and YOLO, this resource by laclouis5 provides standardized tools for object detection evaluation as of February 2026. It enables the calculation of mean Average Precision metrics for bounding box accuracy across diverse computer vision datasets.
Almost 4500 images depicting four types of natural elements, likely for computer vision tasks. The dataset is hosted on Kaggle and is associated with a CNN model. Details on the specific disaster types, collection method, and authorship are not provided in the available metadata.
UVH-26 is a large-scale, India-specific traffic-camera image dataset released by AIM @ IISc for research in intelligent transportation systems and vehicle-related computer vision. The dataset was last updated on December 10, 2025. Models trained on this data reportedly deliver up to 31.5% higher mAP than COCO-pretrained baselines.
3.8 million license plate images featuring over 2.6 million labeled plates from international jurisdictions. This collection provides a large-scale resource for optical character recognition (OCR) tasks across diverse plate formats and alphanumeric styles.
1,000 scanned receipt images across text localization and information extraction categories. The dataset provides ground truth labels for company, address, date, and total amount alongside raw OCR transcriptions.
AgriAISeg provides pixel-level segmentation masks for images of three specific crops: tomato, cabbage, and sesame. The dataset is hosted on Kaggle and is designed for computer vision tasks in agricultural research. Details on the dataset's creator, size, and collection date are not provided in the available metadata.
PaddleDetection is an end-to-end object detection development kit. The dataset likely contains images annotated for object detection tasks. The author, organization, and specific data size are unknown.
PaddleYOLO is a collection of implementations for the YOLO series of object detection models. It includes versions YOLOX, YOLOv5, YOLOv6, and YOLOv7, built using the PaddleDetection framework. The dataset likely contains pre-trained models and associated configuration files for computer vision tasks.
YOLOv6 is a single-stage object detection framework. The description indicates it is dedicated to industrial applications. The dataset's specific contents, such as training images or model weights, are not detailed in the provided metadata.
Wendell0218 created this curated repository of motion and trajectory datasets, which was last updated in March 2026. The collection serves as a multimodal index for researchers seeking motion-related data across various scientific and engineering domains.
This collection aggregates multiple PCB manufacturing datasets into a standardized format for detecting industrial defects. It provides labeled image data across categories such as short circuits, open circuits, and mouse bites, specifically pre-processed for YOLO-based object detection architectures.
The YouTube-8M Segments dataset is an extension of the YouTube-8M dataset for video understanding. It is tagged for tasks like Video Classification and Computer Vision.
A dataset of ancient Tamil script images for optical character recognition and classification tasks. It contains 59 distinct character classes.
A pre-trained Mask R-CNN model published on Kaggle. The dataset likely contains model weights and configuration files for performing instance segmentation tasks. The author, organization, and last update date are unknown.
ResNet50-pretrained is a dataset containing a pre-trained ResNet50 model for image classification, hosted on Kaggle. The model is likely trained on a large-scale image dataset such as ImageNet. Specific details on the model's exact training data, version, and architecture parameters are not provided in the minimal metadata.
ResNet50 is a widely used convolutional neural network architecture with 50 layers. This Kaggle resource provides a pretrained version of the model, likely intended for transfer learning tasks. The specific training data, performance metrics, and fine-tuning details require verification after download.
A collection of videos hosted on Kaggle, likely intended for testing and developing YOLO (You Only Look Once) object detection models. The dataset's specific content, size, and creation details are not provided in the metadata. Its utility for computer vision tasks must be verified after download.
A source of quantum chemical energy calculations for organic molecules computed at the high-level DLPNO-CCSD(T) level of theory. It features paired results using both def2-SVP and def2-TZVP basis sets to facilitate the training of machine learning models for Δ-CCSD(T) energy prediction.
Object Detection Data is a Kaggle-hosted dataset for computer vision tasks. The dataset likely contains images with bounding box annotations for training and evaluating detection models. Its specific scale, source, and creation date are not detailed in the provided metadata.