Vgg16

models/ResNet50/README.md

@@ -0,0 +1,44 @@
+# Infant Cry Classification ResNet50 Model
+
+# Overview
+This repository contains a ResNet50 model for classifying infant cry sounds. The model achieved an accuracy of 84.273% on the test dataset, showcasing its ability to capture intricate features of infant cry patterns.
+
+# Model Architecture
+The ResNet50 architecture is designed to facilitate training of very deep networks. It includes residual blocks that enable the training of deeper networks without the vanishing gradient problem.
+
+
+Model: "resnet50"
+__________________________________________________________________________________________
+Layer (type)                    Output Shape         Param #     Connected to
+==========================================================================================
+input_1 (InputLayer)            [(None, 224, 224, 3) 0
+__________________________________________________________________________________________
+conv1_pad (ZeroPadding2D)       (None, 230, 230, 3)  0           input_1[0][0]
+__________________________________________________________________________________________
+conv1_conv (Conv2D)             (None, 112, 112, 64 9472        conv1_pad[0][0]
+...
+__________________________________________________________________________________________
+dense_3 (Dense)                 (None, 1)            2049        global_average_pooling2d_1[0][0]
+==========================================================================================
+Total params: 23,587,713
+Trainable params: 23,534,593
+Non-trainable params: 53,120
+
+# Dataset
+The model was trained on a diverse dataset containing recordings of infant cry sounds. The dataset includes various cry patterns and non-cry sounds to ensure robust classification.
+
+# Training
+The ResNet50 model was trained using TensorFlow and Keras with an Adam optimizer. The training process involved data augmentation techniques to enhance model generalization. The training accuracy reached 90%, while the validation accuracy reached 88%.
+
+# Evaluation
+The model achieved an accuracy of 84.273% on the test dataset even though dataset is bit imbalanced, highlighting its ability to accurately classify infant cry sounds. The model's precision, recall, and F1-score metrics are commendable.
+
+# Usage
+To use the trained ResNet50 model for inference, you can load the model weights using the provided script:
+
+
+python load_resnet50_model.py --weights path/to/resnet50_weights.h5 --audio path/to/test_audio.wav
+Replace path/to/resnet50_weights.h5 with the path to the saved model weights and path/to/test_audio.wav with the path to the audio file you want to classify.
+
+# Acknowledgments
+I would like to express our gratitude to the Maintainers and data providers who made this project possible.

models/VGG16/README.md

@@ -0,0 +1,40 @@
+# Infant Cry Classification VGG16 Model
+
+# Overview
+This repository contains a VGG16 model for classifying infant cry sounds. The model achieved an accuracy of 84.593% on the test dataset, demonstrating its effectiveness in capturing intricate features of infant cry patterns.
+
+# Model Architecture
+VGG16 is a convolutional neural network architecture that gained popularity for its simplicity and effectiveness. It comprises multiple convolutional and max-pooling layers, followed by fully connected layers.
+
+Model: "vgg16"
+__________________________________________________________________________________________
+Layer (type)                    Output Shape         Param #     Connected to
+==========================================================================================
+input_1 (InputLayer)            [(None, 224, 224, 3) 0
+__________________________________________________________________________________________
+block1_conv1 (Conv2D)           (None, 224, 224, 64) 1792        input_1[0][0]
+...
+__________________________________________________________________________________________
+dense_3 (Dense)                 (None, 1)            513         dense_2[0][0]
+==========================================================================================
+Total params: 138,357,513
+Trainable params: 138,357,513
+Non-trainable params: 0
+
+# Dataset
+The model was trained on a diverse dataset containing recordings of infant cry sounds. The dataset includes various cry patterns and non-cry sounds to ensure robust classification.
+
+# Training
+The VGG16 model was trained using TensorFlow and Keras with an Adam optimizer. The training process involved data augmentation techniques to enhance model generalization. The training accuracy reached 91%, while the validation accuracy reached 89%.
+
+# Evaluation
+The model achieved an accuracy of 84.593% on the test dataset, showcasing its ability to accurately classify infant cry sounds. The model's precision, recall, and F1-score metrics are impressive.
+
+# Usage
+To use the trained VGG16 model for inference, you can load the model weights using the provided script:
+
+python load_vgg16_model.py --weights path/to/vgg16_weights.h5 --audio path/to/test_audio.wav
+Replace path/to/vgg16_weights.h5 with the path to the saved model weights and path/to/test_audio.wav with the path to the audio file you want to classify.
+
+# Acknowledgments
+I would like to express my gratitude to the contributors and data providers who made this project possible.