This research focused on developing a deep learning model for trip demand prediction. The research work explored the effect of demographic, and land-use types on trip demand by society. A new model, titled CityNet, was developed, which captures correlations in trip data by learning the socio-economic and land-use features of city regions. The model predicts the number of incoming and outgoing trips to all city regions and then utilizes these values to estimate trip demand between region pairs. Our experimental results show that CityNet achieves a 56% improved error rate compared to several time-series forecasting techniques and neural network methods, in addition to a 15% improvement compared to two variants of our model.