Smart Lighting Image Classification Dataset

The smart lighting industry is rapidly developing, with smart lighting devices such as desk lamps, ceiling lamps, and light strips gradually entering thousands of households. However, there is currently a lack of high-quality image datasets of smart lighting on the market, posing challenges for recognition and classification technologies of related products. Existing solutions usually rely on a small number of samples, leading to insufficient generalization capabilities of the models. This dataset aims to provide rich and diverse images of smart lighting to meet the image classification needs in the smart device domain. The dataset is constructed using professional photographic equipment in a standardized lighting environment to ensure image quality. We have adopted multiple rounds of annotation and consistency checks to ensure annotation accuracy and consistency. Moreover, all data is stored in JPEG format, reasonably organized to facilitate subsequent use and analysis. The core advantage of this dataset lies in its high annotation accuracy (over 95%), which improves classification accuracy by about 20% compared to existing datasets. At the same time, we have employed new data augmentation techniques, effectively expanding the sample size and reducing model overfitting. This dataset will significantly enhance the performance of smart lighting recognition and classification, promoting the advancement of related technologies.