MDE

Version 1.0.0 (18,9 MB) von Brian Park
Monocular Depth Estimation for indoor environment with MATLAB
0.0
(0)
69 Downloads
Aktualisiert 28. Jul 2023
Lizenz auf GitHub anzeigen

Monocular-Depth-Estimation

View MDE(Monocular Depth Estimation) on File Exchange

Monocular Depth Estimation for indoor environment with MATLAB

This repository contains the MATLAB project which computes depth map from a single image and transforms into point cloud.

Installation

Download ZIP file and run main script

    run('main.mlx');

Run training script only for training network

    run('train_net.mlx');

Dataset not included. Download NYU v2 dataset

Dependencies

MATLAB

  • Deep Learning Toolbox
  • Computer Vision Toolbox

Simulink

  • Robotics System Toolbox
  • Automated Driving Toolbox
  • PackageGazeboPlugin (from Gazebo Simulation Environment Requirements and Limitations )

Device

  • Ubuntu® Ubuntu 20.04.3 LTS (Focal Fossa)
  • CMake 3.16.3, Gazebo 11, and the Gazebo plugin

Usage

  1. Load downloaded dataset and Check
  2. Design the network using DeepNetworkDesigner
  3. Train the network using training option in DeepNetwrokDesigner, and then Save the trained network in the directory.
  4. Load the trained network and test in MATLAB
  5. Load the network in Simulink and Connect the simulation with Gazebo

Zitieren als

Brian Park (2025). MDE (https://github.com/robot-park/Monocular-Depth-Estimation/releases/tag/v1.0.0), GitHub. Abgerufen.

Kompatibilität der MATLAB-Version
Erstellt mit R2023b
Kompatibel mit allen Versionen
Plattform-Kompatibilität
Windows macOS Linux
Tags Tags hinzufügen

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

dataset

function/postprocess

function/preprocess

function/training

function/visualize

_simulation

model/net_MDE

Version Veröffentlicht Versionshinweise
1.0.0

Um Probleme in diesem GitHub Add-On anzuzeigen oder zu melden, besuchen Sie das GitHub Repository.
Um Probleme in diesem GitHub Add-On anzuzeigen oder zu melden, besuchen Sie das GitHub Repository.