Scale-Adaptive KCF Mixed with Deep Feature for Pedestrian Tracking

Pedestrian tracking is an important research content in the field of computer vision. Tracking is achieved by predicting the position of a specific pedestrian in each frame of a video. Pedestrian tracking methods include neural network-based methods and traditional template matching-based methods, such as the SiamRPN (Siamese region proposal network), the DASiamRPN (distractor-aware SiamRPN), and the KCF (kernel correlation filter). The KCF algorithm has no scale-adaptive capability and cannot effectively solve the occlusion problem, and because of many defects of the HOG (histogram of oriented gradient) feature that the KCF uses, the tracking target is easy to lose. For those defects of the KCF algorithm, an improved KCF model, the SKCFMDF (scale-adaptive KCF mixed with deep feature) algorithm was designed. By introducing deep features extracted by a newly designed neural network and by introducing the YOLOv3 (you only look once version 3) object detection algorithm, which was also improved for more accurate detection, the model was able to achieve scale adaptation and to effectively solve the problem of occlusion and defects of the HOG feature. Compared with the original KCF, the success rate of pedestrian tracking under complex conditions was increased by 36%. Compared with the mainstream SiamRPN and DASiamRPN models, it was still able to achieve a small improvement.

Online dictionary learning for car recognition using sparse coding and LARS

<p>The bag of feature method coupled with online dictionary learning is the basis of our car make and model recognition algorithm. By using a sparse coding computing technique named LARS (Least Angle Regression) we learn a dictionary of codewords over a dataset of Square Mapped Gradient feature vectors obtained from a densely sampled narrow patch of the front part of vehicles. We then apply SVMs (Support Vector Machines) and KMeans supervised classification to obtain some promising results.</p>

Gradient Feature Extract for the Quantification of Complex Defects Using Topographic Primal Sketch in Magnetic Flux Leakage

Defect quantification is a very important aspect in nondestructive testing (NDT) as it helps in the analysis and prediction of a structure's integrity and lifespan. In this paper, we propose a gradient feature extraction for the quantification of complex defect using topographic primal sketch (TPS) in magnetic flux leakage (MFL) testing. This method uses four excitation patterns so as to obtain MFL images from experiment; a mean image is then produced, assuming it has 80–90% the properties of all four images. A gradient manipulation is then performed on the mean image using a novel least-squares minimization (LSM) approach, for which, pixels with large gradient values (considered as possible defect pixels) are extracted. These pixels are then mapped so as to get the actual defect geometry/shape within the sample. This map is now traced using a TPS for a precise quantification. Results have shown the ability of the method to extract and quantify defects with high precision given its perimeter, area, and depth. This significantly eliminates errors associated with output analysis as results can be clearly seen, interpreted, and understood.