Table 8.

Methodology and Limitations: Camera-based 3DOR.

	Camera
Methodology	Uses RGB image for object detection and predicts 2D BBoxes, which are inferred to generate 3D BBoxes by re-projection or BBox regression, computationally less expensive compared to other methods
Limitation(s)	Input image does not have depth information, which causes low localization performance and inaccurate object size estimation
Research Gap	CNN architectures for estimating the depth information need to be investigated to improve the detection results