inuwamobarak/african_attire_detector

This repo is made public for my article in Analytics Vidhya. Using deep learning and picture classification, we created a working prototype for fashionistas in detecting African Attires. The objective is to carry out a comprehensive machine-learning demo utilizing a scenario of a real-world issue. # Deep learning for Fashionistas: African Attire Detection This repository contains a deep learning project focused on African attire detection. The project aims to develop a model that can classify eight local African attires, primarily from countries like Nigeria and South Africa. By preserving and utilizing this model, it becomes easier for foreigners and future generations to identify and appreciate the cultural artifacts of African tribes. ## Table of Contents - Problem Statement - Approach - Dataset Description - Installation - Usage - Technologies - Contributing - License - References ## Problem Statement The preservation of African cultures is crucial, especially with the encroachment of civilization and technology. This project aims to leverage AI to combat culture loss and preserve African heritages. By developing deep learning algorithms and smart systems like chatbots, image identifiers, and text-to-speech, we can ensure the preservation and accessibility of African cultures. ## Approach The project utilizes a deep learning algorithm with TensorFlow as the backend framework. The development environment is Colab, and the accuracy of the model is open to further improvement. The goal is to train the model to classify eight African attires accurately. ## Dataset Description The dataset consists of images of African attires gathered from the internet using a web scraping Google Chrome extension. The dataset contains the following details: - 8 classes representing different African tribes: - Adire (from Nigeria) - Idgo (from Nigeria) - Idoma (from Nigeria) - Igala (from Nigeria) - Tiv (from Nigeria) - Tswana-Shweshwe (from South Africa) - Xhosa-South Africa (from South Africa) - Zulu (from South Africa) - Training set: 9784 images belonging to the 8 classes - Validation set: 2579 images belonging to the 8 classes - Total: 12,363 images ## Installation To use the code in this repository, follow these steps: 1. Clone the repository: `git clone http …