This project is the part of my machine learning internship at Unified Mentor. The goal of this project was to develop an image classification system capable of identifying animals in given images. The model should have been able to classify 15 different species of animals.
The orginal dataset consists of 15 folders, each containing images of dimensions 224 x 224 x 3. The classes in the dataset are:
- Bear
- Bird
- Cat
- Cow
- Deer
- Dog
- Dolphin
- Elephant
- Giraffe
- Horse
- Kangaroo
- Lion
- Panda
- Tiger
- Zebra
To achieve accurate classification, I explored various machine learning techniques, with a focus on Neural Networks and Transfer Learning. These methods provided an optimal solution for the required image classification task.
-
app_screenshots/: Contains the screenshots of the image classficiation system. -
classify.py: The source code for the image classification system, including a Gradio-based graphical user interface (GUI). -
Animal_Image_Model_Training.ipynb: Jupyter notebook for model development and training. -
README.md: This file, providing an overview of the project. -
requirements.txt: Dependencies that are required to run the system
A working demo is uploaded in hugginface spaces. Here is the link - https://huggingface.co/spaces/nsk246/Animal_Classify_UnifiedMentor
A trained model (based on the orginal dataset provided) via the notebook is already saved at https://file.io/HwQgcOuBZ12m . To train a new model:
- Run the Python notebook.
- Feel free to customize hyperparameters, batch size, datasets, and augmentation to create your own model.
To run the image classification system, follow these steps:
- Clone the repository.
- Install the required dependencies from
requirements.txt. - Run
classify.py. - Upload test images and run the system via the Gradio interface.
- You can view the predicted Animal name along with the confidence score for the predicted class.