Physics Lab Equipment Image Dataset for Vision-Based Detection

The Physics Laboratory Equipment Image Dataset consists of 5,121 high-quality images covering 27 commonly used physics laboratory instruments, captured under real laboratory conditions. The images exhibit significant variation in viewpoint, scale, background, and illumination, making the dataset suitable for robust deep learning–based object detection and recognition tasks. A total of 8,700 object instances are carefully annotated using bounding boxes and polygon labels to ensure accurate object localization across diverse equipment geometries.All annotations are provided in YOLO-compatible TXT format, containing class identifiers and normalized bounding box coordinates. The dataset is organized into training (70%), validation (20%), and testing (10%) splits, with a standardized image resolution of 640×640 pixels and automatic orientation correction applied during preprocessing. The structured folder layout includes separate images/ and labels/ directories for each split, ensuring ease of use and reproducibility. This dataset serves as a reliable benchmark for physics laboratory equipment detection and supports applications in automated lab management, educational tools, and intelligent laboratory systems.