From 64942202db15730f43eeb5ecd933d95c1aa112b3 Mon Sep 17 00:00:00 2001 From: Vanshika Bisht Date: Sun, 12 May 2024 18:02:14 +0530 Subject: [PATCH 1/3] Update README.md --- .../Model/README.md | 92 ++++++++++--------- 1 file changed, 50 insertions(+), 42 deletions(-) diff --git a/Children vs Adults Classification using DL/Model/README.md b/Children vs Adults Classification using DL/Model/README.md index a780d7c4a..55f0da5dc 100644 --- a/Children vs Adults Classification using DL/Model/README.md +++ b/Children vs Adults Classification using DL/Model/README.md @@ -1,67 +1,75 @@ -# Children vs Adults Classification +# CHILDREN VS ADULT CLASSIFICATION -**PROJECT TITLE** +## ๐ŸŽฏ Goal -Children vs Adults Classification +The main goal of this project is to develop a classification model capable of accurately distinguishing between images of children and adults. The purpose is to explore the performance of different deep learning models specifically tailored for this classification task. -**GOAL** +## ๐Ÿงต Dataset -The main goal of this project is to develop a classification model that can accurately distinguish between images of children and adults. The purpose of the project is to explore the performance of different deep learning models in this specific classification task. +The dataset used for this project can be found [here](https://www.kaggle.com/datasets/die9origephit/children-vs-adults-images). It consists of a collection of labeled images containing children and adults. -**DATASET** - -The dataset used for this project can be found at [link to dataset](https://www.kaggle.com/datasets/die9origephit/children-vs-adults-images). The dataset consists of a collection of images labeled as either children or adults. - -**DESCRIPTION** +## ๐Ÿงพ Description This project aims to build a classification model that can analyze facial features and classify images as either children or adults. By leveraging deep learning models, the project seeks to achieve accurate and reliable classification results. -**WHAT I HAD DONE** +## ๐Ÿงฎ What I had done! -1. Data collection: Gathered a diverse dataset of images containing children and adults. -2. Data preprocessing: Performed necessary preprocessing steps such as resizing, normalization, and augmentation. -3. Model selection: Chose popular deep learning models, including VGG19, ResNet50, InceptionV3, and MobileNetV2, for the classification task. -4. Model training: Trained each model using the labeled dataset and appropriate training configurations. -5. Model evaluation: Evaluated the trained models on a separate test dataset to measure their performance. -6. Comparative analysis: Compared the accuracy and results of each model to determine the best-performing model. +1. Data collection - Gathered a diverse dataset of images containing children and adults. + +2. Data preprocessing - Performed essential preprocessing steps, including resizing, normalization, and augmentation, to prepare the data for training. + +3. Model selection - Chose popular deep learning models, including VGG19, ResNet50, InceptionV3, and MobileNetV2, for the classification task. + +4. Model training - Trained each model using the labelled dataset and appropriate training configurations. + +5. Model evaluation - Evaluated the trained models on a separate test dataset to measure their performance in terms of accuracy and other relevant metrics. + +6. Comparative analysis - Compared the accuracy and results of each model to determine the best-performing model for the task of classifying images into children and adults categories. -**MODELS USED** +## ๐Ÿš€ Models Implemented The following models were used in this project: -1. VGG19 -2. ResNet50 -3. InceptionV3 -4. MobileNetV2 +1. VGG19 - VGG19 is a deep convolutional neural network known for its simplicity and effectiveness in image classification tasks. It consists of 19 layers and has achieved remarkable accuracy in various competitions and benchmarks. + +2. ResNet50 - ResNet50 is part of the ResNet (Residual Network) architecture, featuring residual connections that enable the training of very deep networks. ResNet50 specifically has 50 layers and has demonstrated superior performance in image classification tasks, especially on datasets with a large number of classes. + +3. InceptionV3 - InceptionV3, developed by Google, is famous for its inception module, which allows for efficient use of computational resources by parallelizing operations. It has been widely adopted due to its excellent trade-off between computational efficiency and accuracy, + +4. MobileNetV2 - MobileNetV2 is designed specifically for mobile and embedded vision applications, where computational resources are limited. It utilizes depth-wise separable convolutions to reduce the number of parameters and computations while maintaining high accuracy. -The choice of these models was based on their proven performance in image classification tasks and their varying architectural complexities. This allowed for a comprehensive analysis of different model types. +The choice of these models was based on their proven performance in image classification tasks and their varying architectural complexities, enabling a comprehensive analysis. -**LIBRARIES NEEDED** +## ๐Ÿ“š Libraries Needed The following libraries are required to run this project: + - TensorFlow - An essential deep learning framework offering a flexible ecosystem for building and training neural networks. + + - Keras - A high-level neural networks API, seamlessly integrated with TensorFlow, simplifying the process of building and training deep learning models. + + - Numpy - The fundamental package for scientific computing in Python, providing support for large multi-dimensional arrays and matrices. + + - Matplotlib - A versatile plotting library for Python, enabling visualization of data and model performance with ease. + + - Pandas - A powerful data manipulation and analysis library, facilitating data preprocessing and exploration. -- TensorFlow -- Keras -- numpy -- matplotlib -- pandas - -**VISUALIZATION** - -Comparison Image +## ๐Ÿ“Š Exploratory Data Analysis Results -**ACCURACIES** +![comparison2](https://github.com/vanshikab52/DL-Simplified/assets/148718670/fa02141c-e5c6-41d9-a518-27c091af435d) -The accuracy results obtained for each model on the test dataset are as follows: +## ๐Ÿ“ˆ Performance of the Models based on the Accuracy Scores -- VGG19: 0.73 -- ResNet50: 0.67 -- InceptionV3: 0.74 -- MobileNetV2: 0.79 +- VGG19 - 0.73 +- ResNet50 - 0.67 +- InceptionV3 - 0.74 +- MobileNetV2 - 0.79 -**CONCLUSION** +## ๐Ÿ“ข Conclusion -Based on the accuracy results, MobileNetV2 achieved the highest accuracy of 79% on the test dataset, making it the best-fitted model for this particular project. The other models also performed well but had slightly lower accuracies. +Based on the accuracy results, MobileNetV2 achieved the highest accuracy of 79% on the test dataset, making it the best-fitted model for this particular project. The other models also performed well but had slightly lower accuracies. +This project demonstrates the effectiveness of deep learning models in classifying images of children and adults based on facial features, with potential applications in age estimation or child/adult recognition systems. -This project demonstrates the effectiveness of deep learning models in classifying images of children and adults based on facial features. The findings suggest that MobileNetV2 can reliably classify images in this domain, paving the way for potential applications in age estimation or child/adult recognition systems. +## โœ’๏ธ Your Signature + ##### ๐Ÿ“Œ README.md modified by *Vanshika Bisht* @ GGSoC2024 +[![LinkedIn](https://img.shields.io/badge/linkedin-%230077B5.svg?style=for-the-badge&logo=linkedin&logoColor=white)]( www.linkedin.com/in/vanshika-bisht-a875aa2b7) [![GitHub](https://img.shields.io/badge/github-%23121011.svg?style=for-the-badge&logo=github&logoColor=white)](https://github.com/vanshika52) From 1acd21d7f90e22daeb7b6b50d2309208abab8315 Mon Sep 17 00:00:00 2001 From: Vanshika Bisht Date: Sun, 12 May 2024 20:28:15 +0530 Subject: [PATCH 2/3] Update README.md --- .../Dataset/README.md | 13 +++++++------ 1 file changed, 7 insertions(+), 6 deletions(-) diff --git a/Children vs Adults Classification using DL/Dataset/README.md b/Children vs Adults Classification using DL/Dataset/README.md index 0c914a221..9f70ed41d 100644 --- a/Children vs Adults Classification using DL/Dataset/README.md +++ b/Children vs Adults Classification using DL/Dataset/README.md @@ -1,8 +1,9 @@ -The link for Children vs Adults Classification: https://www.kaggle.com/datasets/die9origephit/children-vs-adults-images +### Dataset Link of Children Vs Adults Classification: https://www.kaggle.com/datasets/die9origephit/children-vs-adults-images -The dataset contains:
-60 test images for children class
-60 train images for children class
-340 test images for adult class
-340 train images for adult class +## โ„น๏ธ About the Data +The dataset comprises of: +- **60** test images for children class. +- **60** train images for children class. +- **340** test images for adult class +- **340** train images for adult class. From 18e14762cefc36077e727f20548e48809440ea4e Mon Sep 17 00:00:00 2001 From: Vanshika Bisht Date: Sun, 12 May 2024 20:30:05 +0530 Subject: [PATCH 3/3] Updated README.md --- Children vs Adults Classification using DL/Model/README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/Children vs Adults Classification using DL/Model/README.md b/Children vs Adults Classification using DL/Model/README.md index 55f0da5dc..bcccbb992 100644 --- a/Children vs Adults Classification using DL/Model/README.md +++ b/Children vs Adults Classification using DL/Model/README.md @@ -72,4 +72,4 @@ This project demonstrates the effectiveness of deep learning models in classifyi ## โœ’๏ธ Your Signature ##### ๐Ÿ“Œ README.md modified by *Vanshika Bisht* @ GGSoC2024 -[![LinkedIn](https://img.shields.io/badge/linkedin-%230077B5.svg?style=for-the-badge&logo=linkedin&logoColor=white)]( www.linkedin.com/in/vanshika-bisht-a875aa2b7) [![GitHub](https://img.shields.io/badge/github-%23121011.svg?style=for-the-badge&logo=github&logoColor=white)](https://github.com/vanshika52) +[![LinkedIn](https://img.shields.io/badge/linkedin-%230077B5.svg?style=for-the-badge&logo=linkedin&logoColor=white)](www.linkedin.com/in/vanshika-bisht-a875aa2b7) [![GitHub](https://img.shields.io/badge/github-%23121011.svg?style=for-the-badge&logo=github&logoColor=white)](https://github.com/vanshikab52)