Code for the Airbus Ship Detection on Kaggle, modified for the Machine Learning for Vision and Multimedia class at Politecnico di Torino, with a focus on Data Augmentation Techniques
- Used schedulerplotter
- Bayesian Data Augmentation
- Fourier Based Data Augmentation
- Gaussian Patch Data Augmentation
- Random Noise Data Augmentation
- Geometric Transformations
The target (ground-truths) are a list of dictionaries, containing three fields
"boxes": tensor([], dtype=torch.float64),
"label": tensor([], dtype=torch.int64),
"image_id": str
This repository used git lfs
for tracking large files, read more about it here
Instructions for Arch Linux
paru -Syu git-lfs
git lfs install