A newer version of the Gradio SDK is available:
5.42.0
title: 🌼 EfficientNetV2B0 Flower Classifier
emoji: 🌸
colorFrom: yellow
colorTo: pink
sdk: gradio
app_file: app.py
pinned: true
🌼 EfficientNetV2B0 Flower Classifier
An elegant and efficient image classifier trained to recognize 5 flower types: daisy, dandelion, roses, sunflowers, and tulips.
Powered by TensorFlow, fine-tuned using EfficientNetV2B0, and deployed with Gradio on Hugging Face Spaces.
Live Demo
👉 Try the app here: Hugging Face Space
Upload a flower image and get the top 5 predictions with confidence scores.
Model Details
- Backbone: EfficientNetV2B0 (
keras.applications) - Framework: TensorFlow 2.x
- Dataset: TensorFlow Flowers (~3,700 images, 5 classes)
- Classes:
daisy,dandelion,roses,sunflowers,tulips - Validation Accuracy: 91.28%
- Training Strategy:
- Stage 1: 5 epochs (base frozen)
- Stage 2: 5 epochs (fine-tuning all layers)
- Preprocessing:
preprocess_input()scaled to [-1, 1]
📓 Training Notebooks
✅ Kaggle: Flower Recognition – Fine-Tuning EfficientNetV2B0
Full training notebook with dataset loading, preprocessing, model building, and evaluation.
⚠️ Colab: (Archived) Training started in Google Colab but was moved to Kaggle due to GPU quota limitations.
You can still view the original Colab notebook here: Colab Fine-Tuning
## 📁 Project Structure
efficientnet-flower-classifier/
├── app.py # Gradio app (entry point)
├── models/
│ └── flower_model.h5 # Trained Keras model
├── requirements.txt
└── README.md
Run Locally
git clone https://github.com/YOUR_USERNAME/8_FlowerRecognition-HF.git
cd 8_FlowerRecognition-HF
pip install -r requirements.txt
python app.py
Dependencies
- tensorflow
- gradio
- numpy
- pillow
Acknowledgments
EfficientNetV2 Paper — Tan & Le