Flower Gender Prediction Model
This model predicts the gender of corn, pumpkin, and ginkgo flowers using a Convolutional Neural Network (CNN).
The purpose is to help researchers and practitioners to identify the gender of flowers more quickly and accurately than manual methods.
Model Details
- Task: Image Classification, Flower Gender Classification
- Framework: TensorFlow/Keras/Flask
- Input: 150x150 images of corn, pumpkin, and ginkgo flowers
- Output: Gender classification (Male/Female)
Dataset
- Flower Dataset: Dataset contains images of corn, pumpkin, ginkgo, rose and daisie
- Amount of data: In total there are 860 flower images, 700 for training, 160 for validation
- Dataset Source: Manually collect a dataset of flower images on google
- Main Dataset: https://drive.google.com/drive/folders/1ttEWEji4B7ykRt9rVIdzm3V7Bjjtzg8R?usp=sharing
- Complementary Dataset: https://drive.google.com/drive/folders/1p5S6YvbriGo7VRZf0WgAT9dHti8hHKu5?usp=sharing
Training
- Model Architecture: CNN with 3 convolutional layers plus a dense layer, a dropout layer, and an output layer
- Optimizer: Adam Optimizer
- Epochs: 50 epochs
Performance
- Accuracy: 92% accuracy and 85% validation accuracy
- Evaluation Methods: Precision, Recall, F1-Score
How to Use
Run app1.py in the terminal, after the application opens press 'choose file' then select the flower image you want to predict,
after that press 'upload', then the prediction results will appear at the bottom of the image.
