Fine-grained image classification is a challenging task in computer vision that involves distinguishing between very similar categories within a broader class. Unlike general image classification, where the goal is to categorize images into high-level categories (e.g., dogs vs. cats), fine-grained classification requires identifying subtle differences between subcategories (e.g., different breeds of birds).
The motivation behind tackling fine-grained image classification stems from its significant potential to enhance various real-world applications. In wildlife monitoring, accurately classifying species can aid in conservation efforts by providing precise data on biodiversity.
In the medical field, fine-grained classification can improve diagnostic accuracy by distinguishing between similar-looking diseases. Despite its importance, fine-grained classification remains challenging due to the high intra-class variability and low inter-class variance, making it an intriguing problem for further research and development.
This project aims to explore state of the art methods on Fine-Grained Classification and experiment with them on Benchmark and evaluation Datasets.
images/the folder where all snapshot for the training and testing logs are stored.
environment.yaml: anaconda environment file, to load the environment.README.md: Markdown text with a brief explanation of the project and the repository structure.
git clone https://github.com/unitn-machine-learning/fine-grained-image_classification.git
cd fine-grained-image_classification
conda env create --name fine_grained_img_classification_env --file environment.yaml
conda activate fine_grained_img_classification_env
- Can be found in this root directory.
Make sure you have the following components installed on your local machine.
- Anaconda or Miniconda
- Python == 3.7
Make sure the dataset is downloaded in the appropriate directory. And the config scripts are appropriately organized.
Our Pre-trained Models and Challenge Dataset Can be Found here:
Distributed under the MIT License. See LICENSE for more information.
👤 Alberto Gabriele Scuderi
- GitHub: Alberto Gabriele Scuderi
- Email: Alberto Gabriele Scuderi
👤 Hafiz Muhammad Ahmed
- GitHub: Hafiz Muhammad Ahmed
- Email: Hafiz Muhammad Ahmed
👤 Julius Heiko Schmidt
- GitHub: Julius Heiko Schmidt
- Email: Julius Heiko Schmidt
👤 Yishak Tadele
- GitHub: Yishak Tadele
- Email: Yishak Tadele
- LinkedIn: Yishak Tadele
Give US a ⭐ if you like this project!