A new design method of an all‐digital automatic gain control easy to implement is described and tested. The discretisation of the instantaneous gain provides a recursive form that merges the exponential function with the level detection. The first proposed circuit is well adapted for a software implementation. It comprises a simple control loop and two multipliers. It does neither contain digital filter nor divider and its setting only requires a loop gain parameter. Some tests were conducted on real digital audio signals with a dynamic range of 80 dB and they validate this approach. Under some conditions, the second circuit can even be simpler since it requires only one multiplier. It is well adapted to an field programmable gate array (FPGA) implementation for high speed signal processing.

In the context of smart cities, monitoring pedestrian and vehicle movements is essential to recognize abnormal events and prevent accidents. The proposed method in this work focuses on analyzing video streams captured from a vertically installed camera, and performing contextual road user detection. The final detection is based on the fusion of the outputs of three different convolutional neural networks. We are simultaneously interested in detecting road users, their motion, and their location respecting the static environment. We use YOLOv4 for object detection, FC-HarDNet for background semantic segmentation, and FlowNet 2.0 for motion detection. FC-HarDNet and YOLOv4 were retrained with our orthophotographs dataset. The last step involves a data fusion module. The presented results show that the method allows one to detect road users, identify the surfaces on which they move, quantify their apparent velocity, and estimate their actual velocity.

The N th‐order Walsh–Hadamard transform on a sliding window is described here by means of FIF filters. Factoring the transfer functions of the filters allows for a synthesis of the transform as a tree of comb filters. The trees built in ascending or descending order of the delay value allow the computation of the coefficients, respectively, in the sequency order and in the Hadamard order. The proposed algorithm requires 2 N −2 additions for each projection. The resources used for a parallel field programmable gate array implementation are presented.

In this paper, we present a deep learning processing flow aimed at Advanced Driving Assistance Systems (ADASs) for urban road users. We use a fine analysis of the optical setup of a fisheye camera and present a detailed procedure to obtain Global Navigation Satellite System (GNSS) coordinates along with the speed of the moving objects. The camera to world transform incorporates the lens distortion function. YOLOv4, re-trained with ortho-photographic fisheye images, provides road user detection. All the information extracted from the image by our system represents a small payload and can easily be broadcast to the road users. The results show that our system is able to properly classify and localize the detected objects in real time, even in low-light-illumination conditions. For an effective observation area of 20 m × 50 m, the error of the localization is in the order of one meter. Although an estimation of the velocities of the detected objects is carried out by offline processing with the FlowNet2 algorithm, the accuracy is quite good, with an error below one meter per second for urban speed range (0 to 15 m/s). Moreover, the almost ortho-photographic configuration of the imaging system ensures that the anonymity of all street users is guaranteed.

Electrical arc fault detector development requires many tests to develop and validate detection algorithms. The use of artificial intelligence or mathematical transformation requires the use of consequential datasets of current signatures corresponding to as many different situations as possible. In addition, one of the main drawbacks is that these experiments take a great deal of time and are often laborious in the laboratory. To overcome these limitations, a virtual test bench based on the modeling of a modular 230 VAC electrical circuit has been developed. The simulated network is composed of different home appliances (resistor, vacuum cleaner, dimmer, etc.) and its configurations are those of single and combined loads. The fault modeled is an electric arc, modeled by active diode switching, which can be inserted at any point of the circuit. This arc model takes into account the random variations in the restrike and arc voltage. All the appliance models are validated by comparing the frequential (harmonic distortion) and temporal (agreement index) signatures of the measured currents in real situations to those obtained by modeling. The results obtained using the model and experiment network show that the current signatures are comparable in both cases. Further, two detection algorithms are tested on those current signatures obtained by the modeling and experimentation. The results are comparable and provide identical detection thresholds.