README.md

Archived. This was a group project / final assessment for CSCI 4155/6505 — Machine Learning at Dalhousie University, part of the Artificial Intelligence & Intelligent Systems certificate (undergraduate/graduate mixed course). It is no longer actively maintained. Note: this project was built on PyTorch Lightning, which has since been superseded by Lightning.

Lepus Classifier

A CNN-based image classifier that distinguishes between two rabbit species using only ~85 training images.

Install · Report Bug · Research Report

Quick Start

1. Install dependencies:

pip install -r requirements.txt

2. Login to Weights & Biases for experiment tracking:

wandb login

3. Train a model:

from src.bootstrap import bootstrap

bootstrap()

Or define a custom architecture:

from src.bootstrap import bootstrap
from src.data.model import BaseModel
import torch

class CustomModel(BaseModel):
    def __init__(self):
        super().__init__()
        self.layers = torch.nn.Sequential(...)
        super().__post_init__()

    def forward(self, x):
        return self.layers(x)

bootstrap(model=CustomModel())

See notebooks/example_bootstrap.ipynb for a Google Colab example.

Project Structure

Path	Description
src/	Training pipeline, data loading, and model definitions
notebooks/	Experiment notebooks and usage examples
resources/data.csv	Dataset manifest (image URLs and labels)
docs/report-docs/lepus-classifier-report.pdf	Final research report
docs/poster.pdf	Project poster
docs/proposals/	Initial project proposals

Experiment Logs

Tracked with Weights & Biases: csci-4155-rabbit-classifier

Agent Skill

npx skills add urmzd/lepus-classifier

License

Apache 2.0