Overall Explanation
====================

Overview
---------

Depth sensing is an important capability for many robotic applications, including mapping, navigation, and obstacle detection.
In the past, depth sensing required specialized stereo or RGB-D cameras.
However, recent advances in deep neural networks (DNNs) have made it possible to infer depth from a single monocular image using a technique called mono depth.
This breakthrough has the potential to significantly reduce the cost and complexity of depth sensing for a wide range of robotic applications.


The MIT FastDepth paper presents one such approach to achieving mono depth using Fully-Convolutional Networks (FCNs).
By training FCNs on large datasets of stereo or RGB-D images, FastDepth is able to learn to predict accurate depth maps from a single image.
This technique is not only more affordable than traditional depth sensing methods, but also more flexible, as it does not require specialized cameras.
With the continued advancement of DNNs, mono depth has the potential to become a ubiquitous tool in the field of robotics, enabling robots to better navigate their environments and interact with the world around them.


As you may have noticed, we have only one depth methods and results on our "*follow along*" step. 


FCN-MobileNet
--------------------

While FCN-MobileNet is primarily designed for semantic segmentation tasks, it can also be used for depth estimation.
Depth estimation is the task of estimating the distance from the camera to objects in the scene.


To perform depth estimation using FCN-MobileNet, the network is trained on a dataset of stereo images, which are pairs of images captured from two slightly different perspectives.
During training, the network learns to predict a dense depth map for each image in the pair.


The advantage of using FCN-MobileNet for depth estimation is that it requires fewer computational resources than other deep neural network architectures commonly used for this task.
This makes it well-suited for applications where computational resources are limited, such as on mobile devices or embedded systems.

Overall, FCN-MobileNet's ability to perform both semantic segmentation and depth estimation tasks with high accuracy and low computational requirements makes it a versatile tool for a wide range of computer vision applications.