IMAGE CLASSIFICATION OF HOUSEHOLD BENEFICIARIES OF DIRECT CASH ASSISTANCE USING EFFICIENTNET IN DKI JAKARTA PROVINCE

This study investigates the application of the EfficientNet architecture for image classification to determine eligible recipients of direct cash assistance among households in Jakarta Province. As government efforts to provide aid to citizens increase, it becomes essential to have a system that can accurately recognize and classify eligible populations. Misallocation of aid remains a prevalent issue, often leading to undeserving individuals receiving assistance, which has detrimental consequences. The primary focus is on leveraging deep learning, specifically EfficientNet, to address these challenges. The dataset used consists of house images categorized into two classes: "Mampu" and "Tidak Mampu," which were collected through personal photography and web scraping from Google. The research aims to develop an algorithm that accurately classifies and analyzes the types and eligibility of residential buildings within the general population. Data collection and processing challenges are addressed to ensure the training of high-quality, representative image datasets. The model has demonstrated a high accuracy rate of approximately 95.03% on the validation data.