In this paper, we present a system that has been developed to facilitate the collection and use of Bike-Sharing Systems data for research, notably to develop and compare bike usage forecasting algorithms. We collected internal and external data for six different European cities and developed a system providing short and long-term predictions of bikes and slots availabilities for bike-sharing stations in real-time. In order to provide the best predictions, we developed and compared the performances of two types of algorithm; the first one is based on the state-of-the-art Random Forest algorithm and the second one is based on Convolutional Neural Networks. Our study demonstrates their applicability, showing better accuracy for short-term predictions with the Random Forest algorithm and better long-term prediction accuracy with the Convolutional Neural Networks algorithm.