Enhancing Monocular 3-D Object Detection Through Data Augmentation Strategies

Data augmentation is a crucial component of machine learning. In 2-D object detection tasks, it can significantly enhance the performance of detectors without increasing the inference cost. Data augmentation methods, such as random translation and random resizing, have become standard practices for 2-D object detectors. However, in monocular 3-D object detection tasks, the data augmentation methods used in 2-D object detection cannot be directly applied due to different representations of object positions. In this study, a method is proposed to migrate a 2-D object detection data enhancement method to monocular 3-D object detection while preserving coordinate and size cues. In addition, we address the sampling bias problem associated with data augmentation in this process. We introduce an unbiased sampling (UB) strategy and several new augmentation methods specifically designed for monocular 3-D object detection. Our proposed method achieves a performance of 20.47% AP3D(IOU = 0.7, car, moderate) on the KITTI dataset and a speed of 45 FPS on RTX 2080Ti GPUs, outperforming all previous monocular methods. The source codes are at: https://github.com/jiayisong/DA3D.