Install Apple Sharp Computer Vision Model

In this computer vision tutorial, we explain how to install and run locally the Apple’s new computer vision model called SHARP on a Linux Ubuntu Computer. SHARP is a model that can generate photorealistic 3D views only on the basis of a single image. Sharp obtains a 3D Guassian representation of a depicted scene. This is done in a less than a second. The YouTube tutorial is given below.

Installation procedure

First, you need to install NVIDIA CUDA toolkit by following this tutorial. After that, open a terminal and type

nvcc --version

Then, run this to update and install Git

sudo apt update && sudo apt upgrade
sudo apt install git

Create a workspace folder

cd ~
mkdir testSharp
cd testSharp

Clone the repository

git clone https://github.com/apple/ml-sharp

Create and activate the Python virtual environment:

cd ml-sharp
python3 -m venv env1
source env1/bin/activate

Create input and output folders

mkdir input
mkdir output

Copy the input image files to the input folder. For example, you can use the image given below (just save it to the input folder)

 Install the required dependencies

pip install -r requirements.txt

Copy the input images to the input folder. And run the predictor

sharp predict -i ./input -o ./output  --render

The outputs will be saved in the output folder. Then, open a 3D viewer that can open ply files to see the output files. You can use this online viewer

https://enter-tainer.github.io/browser-sharp/