The project is all about Military aviation more specifically fighter jet image recoginizer which can identify Military aircraft from the image of the aircraft provided as input.
- FastAi for data preprocessing,training etc.
- Gradio.
- Hugging face space for deployment.
- GitHub page for API
An image classification model from data collection, cleaning, model training, deployment and API integration.
The model can classify 26 different types of Fighter Aircrafts
The types are following:
- General Dynamics F-16 Fighting Falcon aircraft
- Mikoyan MiG-29
- Dassault Rafale
- Lockheed Martin F-22 Raptor
- Mikoyan Mig-31
- Sukhoi Su-27
- Lockheed Martin F-35 Lightning II
- KAI KF-21 Boramae
- Sukhoi Su-57
- Shenyang FC-31 Gyrfalcon
- Boeing F-15EX Eagle II
- Eurofighter Typhoon
- AIDC F-CK-1 Ching-kuo
- Lockheed YF-12
- Mikoyan MiG-31
- McDonnell Douglas F-4 Phantom II
- Grumman F-14 Tomcat
- English Electric Lightning
- Focke-Wulf Fw 190
- Mikoyan-Gurevich MiG-25
- Lockheed P-80 Shooting Star
- Sukhoi Su-34
- Chengdu J-20
- Sukhoi Su-35
- Chengdu J-10
- Mikoyan MiG-35
Data Collection: Downloaded from DuckDuckGo using term name
DataLoader: Used fastai DataBlock API to set up the DataLoader.
Data Augmentation: fastai provides default data augmentation which operates in GPU.
Details can be found in notebooks/data_prep.ipynb
Training: Fine-tuned a resnet34 model for 25 epochs (8 times) and got upto ~84% accuracy.
Data Cleaning: This part took the highest time. Since I collected data from browser, there were many noises. Also, there were images that contained. I cleaned and updated data using fastai ImageClassifierCleaner. I cleaned the data each time after training or finetuning, except for the last time which was the final iteration of the model.
I deployed to model to HuggingFace Spaces Gradio App. The implementation can be found in deployment folder or here.
The deployed model API is integrated here in GitHub Pages Website. Implementation and other details can be found in docs folder.
|
|