This project is a part of the CCV1 Tierli Ahluege Group at Data Science FHNW.
This Repository is our solution to the Competition: Conser-vision Practice Area: Image Classification.
Can you classify the wildlife species that appear in camera trap images collected by conservation researchers?
Welcome to the African jungle! In recent years, automated surveillance systems called camera traps have helped conservationists study and monitor a wide range of ecologies while limiting human interference. Camera traps are triggered by motion or heat, and passively record the behavior of species in the area without significantly disturbing their natural tendencies.
However, camera traps also generate a vast amount of data that quickly exceeds the capacity of humans to sift through. That's where machine learning can help! Advances in computer vision can help automate tasks like species detection and classification, localization, depth estimation, and individual identification so humans can more effectively learn from and protect these ecologies.
In this challenge, we will take a look at object classification for wildlife species. Classifying wildlife is an important step to sort through images, quantify observations, and quickly find those with individual species.
This is a practice competition designed to be accessible to participants at all levels. That makes it a great place to dive into the world of data science competitions and computer vision. Try your hand at image classification and see what animals your model can find!
- Deep Learning
- Computer Vision
- Image Classification
- CNN
- Object Detection
- Explorative Dataanalysis
- Data Visualization
- Python
- PyTorch
- wandb
- numpy
- Pandas
- Megadetector
- To use the best Model with some demo images, use this notebook: Demo Model Notebook for the best model
- To inspect the training of the best model, use this notebook: Training Notebook for the best model
- To inspect the Explorative Dataanalysis for the whole dataset, use this notebook: Explorative Dataanalysis Notebook
- If you only want the models, refer to this folder: Folder with the best model
- Clone this repo (for help see this tutorial).
- Demo files are being kept here
- Raw Data is being kept here
- Explorative Dataanalysis Scripts and Files are being kept here
- Megadetector Scripts and data is being kept here
- Models are being kept here
- Models submissions are being kept here
- Source files for training are being kept here
- Source files for pipeline are being kept here
- open
cmd
cd /your/local/github/repofolder/
pipenv install
- Restart VS Code
- Choose the newly created "tierli_ahluege" Virtual Environment python Interpreter
- open
cmd
cd /your/local/github/repofolder/
pipenv sync