kungfuai / plant-traits-kaggle

Our work toward identifying key traits of plants based on images and geodata. Helping ecologists assess the health of ecosystems!

plant-traits-kaggle

Helping ecologist generalize models for predicting plant traits to better understand the health of ecosystems.

Code for this Kaggle competition: https://www.kaggle.com/competitions/planttraits2024/overview

For additional notes see this doc: https://docs.google.com/document/d/1YLDUVcI2sjkkCSk9zewKPOpY5vFpeMFmkDEsXBXCNCU/edit

Quickstart guide

To get up and running with the code, follow these steps. Make sure you have Docker installed.

1. bin/build.sh

2. Download planttraits2024.zip from the Kaggle comp linked above. Unzip it in data/raw

3. bin/preprocess_data.sh to prepare the data and precompute embeddings

4. For W&B logging, contact Nathan to be added to https://wandb.ai/nathan-mandi/PlantTraits2024. Then, create a .env file with your WANDB_API_KEY

You are set up! You should now be able to run bin/train.sh and other scripts to train models, interact with the code in the repo, etc.

Testing

Most of the modules have tests. Run them with bin/test.sh and pass in any args you would pass to pytest. Ex: bin/test.sh -s tests/data/datasets/test_baseline_dataset.py

potluck template readme

This repo uses Potluck, a machine learning repo template from Kung Fu AI. Here are instructions for use.

Requirements

• [Docker][docker-url]
• [Docker Compose][docker-compose-url]
• [NVIDIA Docker Container Runtime][nvidia-url]

Quick Start

• Run bin/build.sh
• Run bin/preprocess_data.sh

Training

To train the model, all you need to do is run this command:

bin/train.sh

(Note: Please include further instructions if GPU is required!)

Testing

Once the Docker image is built we can run the project's unit tests to verify everything is working. The below command will start a Docker container and execute all unit tests using the pytest framework.

bin/test.sh

If you want to run a test on a specific file or directory (rather than running all the tests in the tests/ directory), you can use the -k flag and list the file path afterwards.

For example, if we specifically wanted to run a test called "test_api", and its file path is as "tests/test_api.py", we can run:

bin/test.sh -k test_api.py

Summary of Commands

The bin/ directory contains basic shell bin that allow us access to common commands on most environments. We're not guaranteed much functionality on any generic machine, so keeping these basic is important.

The most commonly used bin are:

• bin/build.sh - build docker container(s) defined in Dockerfile and compose.yaml
• bin/test.sh - run unit tests defined in tests/
• bin/notebook.sh - instantiate a new jupyter notebook server
• bin/shell.sh - instantiate a new bash terminal inside the container
• bin/train.sh - train a model

Additional bin:

• bin/lint.sh - check code formatting for the project
• bin/setup_environment.sh - sets any build arguments or settings for all containers brought up with docker compose
• bin/up.sh - bring up all containers defined in compose.yaml
• bin/down.sh - stops all containers defined in compose.yaml and removes associated volumes, networks and images

Data Directory

Data organization philosophy from cookiecutter data science

├── 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.