Traffic signs detection has become an important feature of Advanced driving assisting systems and even self-driving cars. In this paper, we present an implementation of a traffic signs detection method on Graphics Processing Units (GPU) under real-time conditions. The proposed model is based on deep convolutional neural networks, a deep learning model used in computer vision applications. The deep convolutional neural networks have recently been used to solve many computer vision tasks successfully. Unlike old techniques, the model is used to detect and identify the traffic signs at the same time without the need for any external modules. To achieve real-time inference, we implement the proposed model on the GPU as a natural choice for the implementation of deep learning-based models. Also, we build large traffic signs detection dataset. The dataset contains 10[Formula: see text]000 images captured from the Chinese roads under real-world factors like lightning, occlusion, complex background, etc. 73 traffic sign classes were considered in this dataset. The evaluation of the proposed model on the proposed dataset shows robust performance in terms of speed and accuracy.

Advanced driver assistance system (ADAS) is one of the most important systems for human assistance. It assists the drivers to control the vehicle by providing essential information about the environment objects. In this paper, we propose a traffic signs recognition application for ADAS. The proposed application is based on the deep learning technique. In particular, we used the convolutional neural networks (CNN) to process the data provided by the system cameras. The proposed CNN was scaled in a way to get a light model size without decreasing the accuracy. The proposed CNN is suitable for embedded implementation while keeping high performance and real-time processing. The evaluation of the proposed CNN on the European dataset results in 99.32% accuracy and 250 FPS of inference speed when implemented on an Nvidia GTX960 GPU. The achieved results proved the efficiency of the scaling technique. It is a very good technique to get a small model size and high performance.

The Graphics processors or GPUs have become in a few years powerful tools for applications that require a massively parallel computing. Currently include the applications in multimedia processing, the engineering science and image processing in real time. They offer many advantages such as acceleration of treatment and down energy consumption from an equivalent CPU power. In this paper, we will show the effectiveness of our approach sobel filter (features extraction) by parallelizing the processing applied to different images with different sizes.

Detecting pedestrians is a challenging problem owing to the motion of the subjects, the camera and the background and to variations in pose, appearance, clothing, illumination and background clutter. The Region Covariance Matrix (RCM) descriptors show experimentally significantly out-performs existing feature sets for pedestrian detection. In this paper, we present an efficient features extraction scheme: the Integral CovReg, inspired from Region Covariance Matrix (RCM) descriptors, combined with SVM classifier for pedestrian detection.

In this paper, we propose a wormhole router architecture for symmetric 3D-mesh Networks-on-Chip (NoCs) with virtual channels. It uses the credit-based flow control mechanism and dimension-order routing XYZ algorithm. With priority-based scheduling, our 3D on-chip communication model can support the management of different levels of quality-of-service. The router is implemented on FPGA device using the Xilinx ISE software. Various designs were synthesized to verify the capability of our router. From the implementation results, the proposed router architecture enables a higher data rate and low latency at a reasonable power and area overheads. Furthermore, we demonstrate an analysis and comparison of the cost and performance results between the 2D and 3D designs.