Demo Inception

This small python script simply demonstrates the third version of the inception network from Google. It captures an image using the webcam of the computer and predicts the label using the neural network. For the fun, it will also use the gtts module to say the result.

It is a small demonstration script that needs further cleaning.

Requirements

Make sure you have openCV (it cannot be installed with pip or conda, need to be installed externally, using homebrew for instance)

On mac, if you install opencv using homebrew and virtual environment, you may need to manually link openCV

Installation

1. Clone the repository

   git clone https://github.com/nperraud/demo-inception.git
   

2. Consider creating a pyhton3 virtual environment. Make sure openCV is accessible from it. To test you can simply use

   python
   import cv2
   print(cv2.__version__)
   

3. Install the following python3 packages

   pip install -r requirements.txt
   

Execution

Simply run

python main.py

For the script to work properly, it needs to capture your keyboard input. If your inputs are not captured, click on the current window/figure.

The first time, the script will automatically download the inception network, which will take some time.

Enjoy!