https://github.com/andrewldesousa/mesh-r-cnn

Group Project for 3D Spatial Learning Practical Course at TUM
https://github.com/andrewldesousa/mesh-r-cnn

3d-deep-learning computervision

Last synced: 10 months ago
JSON representation

Group Project for 3D Spatial Learning Practical Course at TUM

• Host: GitHub
• URL: https://github.com/andrewldesousa/mesh-r-cnn
• Owner: andrewldesousa
• Created: 2020-12-11T14:01:52.000Z (over 5 years ago)
• Default Branch: master
• Last Pushed: 2021-03-29T10:49:20.000Z (about 5 years ago)
• Last Synced: 2025-03-17T15:44:42.121Z (about 1 year ago)
• Topics: 3d-deep-learning, computervision
• Language: Python
• Homepage:
• Size: 20.9 MB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
# Mesh R-CNN Playground - Backend

This repository contains a web application for interacting with Mesh R-CNN and our additions to the model. The API has been developed using the FastAPI library.

![Mesh R-CNN Archiecture](docs/mesh_arch.png)

For more information regarding the original Mesh R-CNN work, visit this [link](https://arxiv.org/abs/1906.02739)

## Usage

### Installing dependencies

To install the dependencies for the application, use the requirements.txt file.

`pip install -r requirements.txt` 

### Running the application

For running the app with hot reload functionality.

`uvicorn api.main:app --reload --host 0.0.0.0`

If you don't need hot reload functionality, then run

`uvicorn api.main:app --host 0.0.0.0`

## Functionality

After cloning this repository and setting up the dependencies propery (with requirements.txt), you are able to interact with the model. The API returns everything with the original Mesh R-CNN model plus more. What gets returned are predictions for bounding box, classification, voxel representation, mesh representation, and textured mesh representations.