Kaggle — Indian Birds Classification
Public notebook containing the dataset preparation, EfficientNetB3 training, evaluation, and TensorFlow Lite conversion.
Projects / Indian Birds classification model
Indian Birds classification model
Transfer-learning image classifier for 25 Indian bird species, fine-tuned with EfficientNetB3 and exported to TensorFlow Lite.
machine-learningcomputer-visiontensorflowidcamp
An image-classification experiment completed during the intermediate machine-learning learning phase.
Approach
- Prepared a 25-class Indian bird image dataset with separate training, validation, and test splits.
- Used EfficientNetB3 for transfer learning, then fine-tuned the network for the bird classes.
- Exported both transfer-learning and fine-tuned models to TensorFlow Lite.
The public notebook reports 98.3% accuracy for the final fine-tuned evaluation run. The old portfolio rounded that result to 99%, so this record keeps the more precise public notebook value.
Evidence
Links and references.
Open link ↗
Atlas context
Nearby in the Atlas.
See how this role connects to the projects, case studies, notes, and ideas around it.
overview · 72%