IndiaScene365

This dataset possesses around 3000 high-resolution images with bounding box annotations, captured under daylight, low light, rain and, fog in unstructured driving conditions suitable for Indian traffic scenes. In contrast to other datasets focusing on adverse weather conditions, our collection offers a larger number of annotated and unpaired images, along with an expanded set of labels. This approach allows for the capture of both unstructured and mixed structured-unstructured traffic scenarios. The Indiascene365 dataset has been created to fulfil this requirement, providing a thorough image specifically designed for object detection in Indian traffic environments. This dataset comprises images captured using both high-resolution smartphone cameras and vehicle-mounted front cameras. The images are labeled using an open source roboflow, freely available tool, along with the resulting annotation files to label images for training various deep learning algorithms. Indiascene365 provides a complete representation of both rural and urban Indian traffic scenes, containing over 3000 images encompassing 34 object classes