Men-Women Dataset for Binary Image Classification

This dataset contains images used in the monograph titled Zastosowanie wybranych metod uczenia głębokiego w wizji komputerowej (Application of Selected Deep Learning Methods in Computer Vision) to build classification models. The dataset was prepared based on the original Human Images Dataset – Men and Women available on Kaggle (https://www.kaggle.com/datasets/snmahsa/human-images-dataset-men-and-women). From the original collection, 800 images belonging to each class (man, woman) were used (1,600 images in total). Example images are shown in this Figure: https://drive.google.com/file/d/1byMQyluFyCkNWVsFMbmPzpbfUW-SkZZz/view?usp=drive_link. The images were scaled while maintaining their aspect ratio so that the longer side is equal to 300 pixels. The set was randomly divided into a training part (50% of the full set), validation and test part (each accounted for 25% of the full set). As a result, the training part contains 800 files, validation part – 400 and test part – 400. The dataset can be used to build models for image classification.