The surging demand for sustainable agriculture has accelerated the adoption of indoor vertical farming as a pragmatic solution. Lettuce, a cornerstone crop in this context, assumes significant importance. Accurate forecasting of lettuce yield is indispensable for optimizing resource allocation and ensuring a steady supply. Most existing models used either environmental data or images to predict yield predictions, which could be erroneous for complex systems. This study aims to improve the accuracy of yield prediction in indoor farming settings with a hybrid model. First, we applied the feedforward neural network and random forest models for yield prediction, leveraging data from environmental sensors, cultivation practices, and historical yield records. Then, a convolutional neural network model is tailored to forecast yield using image data captured by RGBD cameras. Based on our results, we found reasonable accuracy in terms of RMSE and MAE, which range between 10-25 gm and 28-49 g, respectively. By amalgamating these diverse models, we aim to elevate yield prediction accuracy. It’s hypothesized that the proposed hybrid model would outperform individual approaches, offering invaluable insights for indoor vertical farming operations decision-making.