Datasets
A versatile collection of high-quality datasets designed to support a wide range of machine learning development tasks, from training and validation to benchmarking across various domains and methods.
Filters
Task
Data Type
Subject area: Quality assessment
Description: HyperNut is a hyperspectral dataset for unsupervised defect detection and segmentation in nuts. It contains visible and near-infrared (VIS-NIR) hyperspectral images of almonds and pistachios, captured in the 400–1000 nm wavelength range.
Dataset: Link
Purpose: Hyperspectral analysis, Anomaly detection
Resolution: 1024x1024
No. Classes: 2 (Almonds, Pistachios)
Samples: Train: 218 images, Test: 144 images
Subject area: Manufacturing
Description: This is a dataset for dense small-object counting in industrial inspection settings, designed to benchmark both exemplar-based few-shot counting and text-guided object counting. It focuses on challenging manufacturing scenarios with small, overlapping, densely packed, and visually similar objects, specifically screws and nuts.
Dataset: Link
Purpose: Object detection and Counting
Resolution: 3024x4032
No. Classes: 2 (Screws, Nuts)
Samples: Train: 1000 images (80,000 samples), Test: 200 images (20,000 samples)
Subject area: Manufacturing
Description: This dataset is collected to train a YOLO model to detect and count different types of packages such as plastic bottles, glass bottles, cans, tubes, jars and cartons in a production line.
Dataset: Link
Purpose: Detect and count packages in a production line.
Resolution: Multi resolutions
No. Classes: 6(plastic bottles, glass bottles, cans, tubes, jars, cartons)
Samples: Train: 9086 samples (801 images), Validation: 445 samples (118 images)
Subject area: Food Industry
Description: This dataset contains high resolution images of white and brown eggs. Egg labels and their locations in each image is provided in text files. The images and their labels are provided in YOLO format that can be used easily for training segmentation and classification models.
Dataset: Link
Purpose: Can be used for developing egg segmentation, classification, and detection models. Moreover, sizes of eggs can be estimated properly which can be used in sorting lines.
Resolution: 3024x4032
No. Classes: 2(white/brown eggs)
Samples: Train:393 egg samples (50 images), Validation: 30 egg samples (2 images)
Subject area: Food Industry
Description: This dataset contains high resolution images of white and brown eggs. Egg labels and their locations in each image is provided in text files. The images and their labels are provided in YOLO format that can be easily used for training detection and classification models.
Dataset: Link
Purpose: Can be used for developing egg detector, classification, and counter models.
Resolution: 3024x4032
No. Classes: 2(white/brown eggs)
Samples: Train:393 egg samples (50 images), Validation: 30 egg samples (2 images)
