Drawer Handle Detection Dataset

In the current retail e-commerce industry, automated identification and handling of products are important means to improve efficiency. However, many existing object detection technologies have insufficient identification accuracy in complex environments, especially in the case of diverse product forms. Our dataset aims to address the recognition accuracy issue in object detection by providing high-quality images of drawer handles and their annotations. The dataset contains 5000 annotated images of drawer handles, captured in high resolution and in diverse environments. During data collection, professional photography equipment was used to shoot under both natural and artificial light to ensure image quality. On the quality control side, the annotations were subjected to multiple rounds of labeling and consistency checks to ensure their accuracy. Data is stored in JPEG format, organized as JSON files containing images and corresponding annotations.