ID: 5d89f4faa0171346e315e541

Map to Satellite GAN

by Deepak Mangla

Automatically converts maps to satellite images with the generative AI.


Contains Dockerfile Ā This model contains Dockerfile.


License: Private

Tags: CycleGAN Map to Satellite Training Data Generation

Ā Model stats and performance
Dataset Used View
Framework PyTorch
OS Used Linux
Publication View
Inference time in seconds per sample.

Screenshots


What is it?

Automatically converts maps to satellite images  with the generative AI.

How to use?

Using pip Prerequisites - cuda 10.1 and python 3.6

  • Install all requirements with command pip install -r requirements.txt.
  • Put one or more images of maps in 'Input' folder. (Initially sample images are provided for testing purpose).
  • Run python src/run.py -- input Input --output Output

Using docker Prerequisites - Docker > 19.03

  • Build docker image and run docker container - docker build -t Map2Sat -f Dockerfile . && docker run -it Map2Sat
  • Run python src/run.py -- input Input --output Output .

Voila ! Maps have been converted to Satellite images and have been stored in 'Output' directory.

Stats:

CPU - 0.96 s GPU - 0.036 s

Tested on

  • python 3.5
  • ubuntu 16.04
  • cuda 10.1

Author View Profile


Deepak Mangla
Faridabad, India
Level 25 34010 XP

1268 Profile
Views

AI wizard who thinks with first principles

I am Deep Learning (Computer Vision) Engineer with specialization in enabling AI on the resource-constrained edge devices. I work on quantizations, pruning and efficient designing and deployment of DL networks. Read my article about Winograd - https://blog.usejournal.com/understanding-winograd-fast-convolution-a75458744ff

User Reviews



0 total ratings

Model has not been reviewed yet.

More by this user | Show All



Also checkout...