There are many ways to estimate depth e.g.
- Structural light based solutions
- Time-of-flight (ToF) cameras
- Stereo cameras
- Computer Vision image based methods
My problem statement is to find the depth of under vehicle portion in real-time. I've tried Intel RealSense Depth Camera, which is a structural light based solution, but it's too sensitive to light conditions, so doesn't give good results. On the other hand, LiDAR is too slow. I haven't tried Zed camera, Kinect or any ToF cameras.
I'm confused whether I should opt for hardware solution or fine-tune a deep learning model for my use-case. I want the depth estimation in real-time for a moving vehicle.