<p>Palm oil, known for its high energy density and wide availability, is a promising resource for sustainable biodiesel production. Biodiesel, derived from renewable sources such as vegetable oils or animal fats, emits fewer pollutants and contributes to reducing environmental issues related to air pollution and climate change. This paper introduces an intelligent, optimized system for estimating biodiesel yield using IoT-based imagery and machine learning (ML) techniques. The proposed system operates in two main phases, which are palm-oil tree detection and biodiesel yield prediction. In the initial phase, palm-oil trees are identified and counted based on input images using the YOLOv9 object detection algorithm. To assess YOLOv9 in various configurations versusYOLOv8, three experiments were carried out. With high-resolution input, YOLOv9 produced the best results with 100% precision and recall and 99.5% mAP50-95. In the second phase, biodiesel yield is predicted using an optimized gradient boosting model based on environmental variables like temperature, humidity, rainfall, and wind speed. With an <i>R</i>² value of 0.98 and RMSE and MSE values close to zero, the system exhibits high inference quality and a fast inference time of 0.00297&#xa0;s. The efficacy of the system under various environmental conditions was confirmed by a real-world case study which also confirmed that the system can accurately estimate the production of biodiesel. This system shows how ML and IoT integration can improve the efficiency of biodiesel production providing a scalable and dependable solution for the development of sustainable energy.</p>