# TensorBird

A project aiming to perform auto identification of bird species from pictures taken by [PiCameraTrap](https://forge.ortion.xyz/UncleSamulus/PiCameraTrap)
using machine learning.

## Inspiration

I have been inprired by following ressource : 
* [darenjhsu birdid repository](https://github.com/darrenjhsu/birdid)
* [This A.I. Birdwatcher Lets You ‘See’ Through the Eyes of a Machine](https://today.duke.edu/2019/10/ai-birdwatcher-lets-you-see-through-eyes-machine)
* [Is Machine Learning for the Birds?](https://www.openhealthnews.com/story/2019-01-24/machine-learning-birds)
* [Image Classification of bird species using deep learning with PyTorch, Captum and ONNX](https://towardsdatascience.com/adventures-in-pytorch-image-classification-with-caltech-birds-200-part-1-the-dataset-6e5433e9897c)
* [Github Repository of previous article](https://github.com/ecm200/caltech_birds)


## Environmnent Set Up

### Create and Activate virtual env

```bash
python -m venv tb-venv
source tb-venv/bin/activate
```

### Install Tensorflow in container

```bash
docker pull tensorflow/tensorflow:latest  # Download latest stable image
docker run -it -p 8888:8888 tensorflow/tensorflow:latest-jupyter  # Start Jupyter server 
```

### Install Tensorflow using pip

```bash
python -m pip install tensorflow
```

### Launch Jupyter Notebook

#### Make the virtual env accessible from Jupyter Notebook

```bash
pip install ipykernel
python -m ipykernel install --user --name=tb-venv
```

#### Launch Jupyter

```bash
jupyter notebook
```



## Run TF Lite

### Trouble shooting

For tensorflow Lite to work on virtual env, we have to install the following package.
```bash
sudo apt install libatlas-base-dev
```