Crop yield forecasting plays a crucial role in improving agricultural planning, food security, and economic stability. Traditional yield prediction methods rely heavily on historical records, weather data, and manual field inspections, which can be time-consuming and less accurate. This project proposes an image-based crop yield forecasting system that uses ground-level crop images captured through mobile cameras or drones instead of satellite imagery. The proposed system analyzes plant characteristics such as leaf color, texture, canopy density, and growth patterns to estimate crop productivity. A deep learning model based on Convolutional Neural Networks (CNN) is trained using labeled crop datasets to identify plant health conditions and growth stages. The system offers a cost-effective and scalable solution for farmers and agricultural planners. The predicted results help farmers make informed decisions related to irrigation, fertilizer usage, and harvesting schedules. Overall, the proposed image-based yield forecasting system improves prediction accuracy and supports sustainable agricultural practices