Social assistance is one of the government programs aimed at improving the welfare of underprivileged communities. The process of determining social assistance recipients is still largely carried out manually, which may lead to subjectivity, inaccurate targeting, and relatively long decision-making processes. This research develops an automatic decision support system for determining the eligibility of social assistance recipients based on house images using Deep Learning and the Analytical Hierarchy Process (AHP). The system design includes the development of a house image classification model using the MobileNetV2 architecture, the development of a web-based system using the Laravel framework, and the integration of classification results with the AHP method. MobileNetV2 is used to classify house conditions into Eligible and Not Eligible categories along with their confidence scores. The confidence scores are then converted into a 1–10 rating scale and used as the House Condition criterion in the AHP calculation, along with income, occupation, number of dependents, and home ownership. This research uses 300 house images as the dataset for model training and testing. The evaluation results show that the MobileNetV2 model achieved an accuracy of 74.00%. The developed system is able to automatically integrate house image classification with AHP calculations, resulting in more objective, consistent, and accurately targeted recommendations for social assistance recipients, while assisting village governments in the process of determining eligible social assistance recipients.