Fruit Classifier

Fruit Edibility AI Jetson Nano

See if I like a certain fruit or not.

In the project, I've listed the fruits that I despise, the ones I love, and the ones I've never tried. Based on that information, the AI return whether I should eat it or not. Just to be clear, the AI is not very accurate as I have not trained it long enough. In the future I'm planning on training even more and making it return the fruits instead.

Prerequisites

  1. Jetson nano
  2. Python installed
  3. USB webcam (For live use)
  4. Jetson-inference (Download below)

Setup

Install CMake

sudo apt-get update, sudo apt-get install git cmake

Install and Clone Jetson Inference

git clone --recursive https://github.com/dusty-nv/jetson-inference, cd jetson-inference, git submodule update --init

Install Python Libraries

sudo apt-get install libpython3-dev python3-numpy

Run CMake

mkdir build, cd build, cmake ../ Make sure googlenet and resnet-18 is selected(will have a star next to it). Others models are optional. No need to install Pytorch.

More Installations

Make sure your still in the build folder

sudo make install, sudo ldconfigRun

Download/Clone this project

Install it manually or with git clone git clone https://github.com/sqwatato/Fruit-Edibility-AI-Jetson-Nano.git

Run AI

Image Input

Replace <resnet_dir> with the path of the directory of the resnet18.onnx file Replace <label_dir> with the path of the directory of the labels.txt file Replace <img_path> with the path of the input image Replace <output_name> with the name of the output image you would like. Ex "fruit.jpg" (.jpg already in command)

`python3 fruit.py --model=/resnet18.onnx --input