Implementing Artificial Neural networks (ANN) for predictive modeling in injection molding encounters notable obstacles due to the considerable volume of training data needed. Transfer learning (TL) provides a viable answer by utilizing knowledge from one dataset to improve the performance of a model on another dataset. The objective of this study is to utilize TL in order to forecast the energy consumption and product weight in injection molding by employing ANN. To acquire energy consumption statistics, it is necessary to conduct experiments directly. Thus, this work will employ complete factorial Design of Experiments (DoE) to acquire a dataset that is both resilient and suitable for training, validation, and testing purposes. When it comes to estimating weight and energy consumption for various materials, studies have shown that the use of transfer learning approach greatly speeds up the learning process. Empirical evidence has demonstrated that employing transfer learning technique significantly improves the coefficient of determination (R2) of the ANN model, reaching values of 93,53% in predicting product weight of 98.87% in energy consumption and 96,25% in total. The deliberate freezing of layers seems to have a more crucial role in enhancing the efficiency and accuracy of the model.