This project helps us in predicting distances of the specified objects in the image from the camera.
This Project uses encoder decoder model for predicting the depth map of the image. The model is trained on NYU V2 dataset which has depth images as ground truth.
The model is still under training.
The following are the results from training the model for 18 epoches
The object detection is still under development and training. (object detection is implemented using single shot detection because of its high FPS in videos)