SynthBRSet (Synthetic Bike Rotation Dataset)

3D Computer Graphics is leveraged to generate a large and diverse dataset for training bike rotation estimators in bike parking assessment. By using 3D graphics software (Blender), the algorithm is able to accurately annotate the rotations of bikes with respect to the parking spot area in two axes y and z , which is crucial for training models for visual object-to-spot rotation estimation. Additionally, the ease of building the algorithm in Python made the generated dataset diverse with a wide range of variations in terms of parking space, lighting conditions, backgrounds, material textures, and colors, as well as objects and camera angles, to improve the generalization of the trained model. Overall, the use of 3D computer graphics allows for the efficient and precise generation of visual data for this task as well as for many potential tasks in computer vision.