Multiple Class Flower Image Classification using Keras
USAGE For training model : python3 training_model --dataset training_set --model trained_model --plot plot For predicting image : python3 predict.py --dataset training_set --model trained_model --image test_set/rose1
The model have been trained using Keras Library.
The architecture of the neural network used here is commonly known as LeNet architecture which is depiced as below INPUT => CONV => RELU => POOL => CONV => RELU => POOL => FC => RELU => FC
For training this model, the most optimized No of Epochs were 25 and the Batch Size of 32 with the Adam Optimizer with the Initial Learning Rate of 1e-3.
The pre-trained model here has the following accuracy/loss which is also shown in the figure plot.png Training Accuracy - 0.9057 Validation Accuracy - 0.8864 Training Loss - 0.2225 Validation Loss - 0.2838