Getting Started

An image classification system is a technology that analyzes and categorizes images based on their visual content. It uses machine learning algorithms to recognize patterns, features, and characteristics within images, enabling it to identify and assign predefined labels or categories to new, unseen images. This system typically involves a training phase where it learns from labeled images to build a model capable of recognizing various objects, scenes, or patterns. Once trained, the system can accurately classify images by assigning them to the appropriate predefined categories or labels.

1. Install library depencies pip install -r requirements.txt
2. Download image dataset from gdrive link in dataset_gambar.txt
3. To make and train the model, run all cell in klasifikasi.ipynb
4. if you want to use the model, download .tflite file

A recipe recommendation system API is a dynamic tool that leverages algorithms and user preferences to suggest personalized cooking ideas. By analyzing user interactions, such as past recipe choices, dietary restrictions, and ingredient preferences, the system generates tailored recommendations to enhance the culinary experience. This API streamlines the process of discovering new recipes, providing users with a curated selection that aligns with their unique tastes and dietary needs. With the ability to adapt and learn from user feedback, the system continually refines its suggestions, ensuring an ever-evolving and enjoyable culinary journey for users.

1.Install library depencies pip install -r requirements.txt
2.Setup database configuration
3.run this command in terminals python -m uvicorn main:app --reload

Pre-trained model: MobileNet Version 2

MobileNetV2 is a pre-trained convolutional neural network architecture designed for efficient and high-performance image classification tasks. It's an improvement over its predecessor, MobileNetV1, optimized for mobile and embedded vision applications. MobileNetV2 introduces inverted residuals and linear bottlenecks, resulting in faster inference speed, lower latency, and improved accuracy compared to the earlier version. Its architecture employs a lightweight depthwise separable convolution, reducing computational complexity while maintaining impressive accuracy on various image recognition tasks. This pre-trained model is widely used as a feature extractor or a base for transfer learning in applications demanding real-time image analysis, such as object recognition in mobile devices, due to its balance between efficiency and accuracy.