Ricania Simulans Detection using CNN

This project use of Convolutional Neural Networks (CNNs) to detect vampire butterflies from images. It involves data preprocessing, model training, and object detection, utilizing a dataset with positive (vampire butterflies) and negative (other butterfly species) samples. The dataset contains subfolders for both positive and negative images, and the preprocessing scripts resize, normalize, and augment these images. The model creation script defines and trains a CNN model using the preprocessed images. Finally, the trained model classifies new images and video frames, offering real-time object detection and video processing functionality.

Model

model = Sequential([

    # Convolutional layers
    Conv2D(filters=32, kernel_size=(3,3), activation='relu', input_shape=(32,32,3)),
    MaxPooling2D(pool_size=(2,2)),
    Dropout(0.1),

    Conv2D(filters=64, kernel_size=(3,3), activation='relu'),
    MaxPooling2D(pool_size=(2,2)),
    Dropout(0.1),

    Conv2D(filters=32, kernel_size=(3,3), activation='relu'),
    MaxPooling2D(pool_size=(2,2)),
    Dropout(0.1),

    # Flatting the layer
    Flatten(),

    # Fully connected layers 
    Dense(units=256, activation='relu'),
    Dropout(0.3),

    Dense(units=128, activation='relu'),
    Dropout(0.3),

    Dense(units=64, activation='relu'),
    Dropout(0.3),
    Dense(units=1, activation='sigmoid')
])

Results