"Custom ImageNet-300 Dataset"

"The Custom ImageNet-300 dataset is a specialized, 300-class subset of the ImageNet (ILSVRC) database, curated for balanced image classification and computer vision research. Comprising 64,904 high-quality JPEG images, the dataset features a standardized resolution of 256x256 pixels across all entries. This collection addresses the need for a diverse yet computationally efficient alternative to the full ImageNet-1K dataset, maintaining a robust class distribution with an average of 216 images per category. Validated through a rigorous automated pipeline to ensure zero corruption, it is particularly suited for benchmarking deep learning models, studying class imbalance, and rapid prototyping of Convolutional Neural Networks (CNNs) in resource-constrained environments."