Deep convolutional neural networks provide state-of-the- art results for image classi cation tasks [1]. Due to the high amount of oating point operations, their implementation in embedded systems is still a challenge, but the rewards in case of success are signi cant. Embedded systems based on FPGA provide a much more ecient solution in terms of power, size and cost when compared with the alternatives (GPUs, workstations). This work presents an ongoing re- search aiming at developing new design methods capable of facilitating the integration of neural networks in image processing applications executing in FPGA. It has been shown [2] that L1 regularization can be used during the training phase of neural networks to reduce the number of oating point operations in multi-layer perceptrons. In this work we further analyze the impact of L1 regulariza- tion in other kinds of neural networks and conclude that pre-processing the data with convolutional layers in the FPGA improve not only the accuracy of the system but also allows for further reduction in oating point opera-tions in the subsequent fully connected layers.