bitjockey42/ikapati-research

ikapati

Plant disease detection using TensorFlow 2.0.

Project Organization

├── LICENSE
├── Makefile           <- Makefile with commands like `make data` or `make train`
├── README.md          <- The top-level README for developers using this project.
├── data
│   ├── external       <- Data from third party sources.
│   ├── interim        <- Intermediate data that has been transformed.
│   ├── processed      <- The final, canonical data sets for modeling.
│   └── raw            <- The original, immutable data dump.
│
├── docs               <- A default Sphinx project; see sphinx-doc.org for details
│
├── models             <- Trained and serialized models, model predictions, or model summaries
│
├── notebooks          <- Jupyter notebooks. Naming convention is a number (for ordering),
│                         the creator's initials, and a short `-` delimited description, e.g.
│                         `1.0-jqp-initial-data-exploration`.
│
├── references         <- Data dictionaries, manuals, and all other explanatory materials.
│
├── reports            <- Generated analysis as HTML, PDF, LaTeX, etc.
│   └── figures        <- Generated graphics and figures to be used in reporting
│
├── requirements.txt   <- The requirements file for reproducing the analysis environment, e.g.
│                         generated with `pip freeze > requirements.txt`
│
├── setup.py           <- makes project pip installable (pip install -e .) so src can be imported
├── ikapati            <- Source code for use in this project.
│   ├── __init__.py    <- Makes src a Python module
│   │
│   ├── data           <- Scripts to download or generate data
│   │   └── make_dataset.py
│   │
│   ├── features       <- Scripts to turn raw data into features for modeling
│   │   └── build_features.py
│   │
│   ├── models         <- Scripts to train models and then use trained models to make
│   │   │                 predictions
│   │   ├── predict_model.py
│   │   └── train_model.py
│   │
│   └── visualization  <- Scripts to create exploratory and results oriented visualizations
│       └── visualize.py
│
└── tox.ini            <- tox file with settings for running tox; see tox.testrun.org

---

Project based on the cookiecutter data science project template. #cookiecutterdatascience

Setup with tensorman

Tensorman is a tool created by the developers of Pop!_OS for managing TensorFlow toolchains. It runs TensorFlow inside docker containers.

Assuming the GPU requirements are satisfied, bring up a container as root:

tensorman run -p 8888:8888 --gpu --python3 --jupyter --root --name ikapati_build bash

Inside that container:

# apt-get install -y python3-venv
pip install poetry

Disable virtualenv creation by poetry since we're running it inside a docker container:

poetry config virtualenvs.create false

This creates a configuration file under .config/pypoetry/, which will be created in this directory since it's mounted in the container.

[virtualenvs]
create = false

Upgrade pip:

# Poetry runs a really older version of pip, which causes installation to fail
poetry run pip install pip==20.0.2
# Install the dependencies
poetry install

In another terminal window on the host, fix the permissions on .config and .cache and save the container as an image:

sudo chown -R $USER {.config,.cache} && sudo chgrp -R $USER {.config,.cache}
tensorman save ikapati_build ikapati

Exit the container, then run a new one as a regular user:

tensorman =ikapati run --gpu --name ikapati_dev bash

For jupyter:

tensorman =ikapati run -p 8888:8888 --gpu --python3 --jupyter bash

Then inside the container:

jupyter lab --ip=0.0.0.0 --no-browser

Setup with poetry

poetry shell
# necessary otherwise dependency resolution will fail
pip install -U pip

Update dependencies and generate new poetry.lock:

poetry update

Install:

poetry install

Setup with conda

With conda, you can install from the environment.yml file:

conda env create --file environment.yml

Alternatively:

conda install pip pandas numpy
conda install -c conda-forge pillow tensorflow-gpu seaborn scikit-learn

Then install requirements:

conda install pip
pip install -r requirements.txt

These environment variables need to be set:

SM_HOSTS=["algo-1","algo-2"]
CUDA_VISIBLE_DEVICES=0

To set these inside the conda env:

# Make sure you're in the env
conda activate ikapati

# Set vars
conda env config vars set CUDA_VISIBLE_DEVICES=0
conda env config vars set SM_HOSTS='["algo-1","algo-2"]'

# Re-activate env to load new variables
conda activate ikapati

# Look at list of set env vars
conda env config vars list

Running

python src/data/make_dataset.py data/fashion_test
python src/models/train_model.py --epochs 20 --batch_size 64 --model_dir models --train data/fashion_test