Technique Development and Applications of Deep Learning for Advanced Driver Assistance Systems(2/2)

Recently, deep learning (DL) or convolutional neural network (CNN) has become the hottest topic ininformation technology due the big success in the applications of artificial intelligence and patternrecognition. Basically, CNN is just revolved from the 90’s traditional artificial neural network; however, itadds the convolutional layers to acquire best features instead of manual features to make the classificationand recognition dramatically improved. In 2016, we processed a face recognition study based on a CNN.Comparing with the traditional approaches for human face recognition: Eigenface based on the principalcomponent analysis (PCA), Fisherface based on the linear discrimination analysis (LDA), support vectormachine (SVM), and adaptive boosting (AdaBoost), the CNN obtained obvious improvement; the recognitionrate was upgraded from about 70 % to 99 %.From 2001 to now, we have studied the visual detection techniques for the advanced driver assistancesystem (ADAS). In this period, we have completed 12 software techniques: 1.lane departure warning (LDW),2.forward collision warning (FCW), 3.blind spot detection (BSD), 4.pedestrian collision warning (PCW),5.traffic sign and signal recognition (TSSR), 6.surrounding top-view monitor (STM), 7.wide-scopicsurrounding top-view monitor and detection (WSTD), 8.image-based parking guiding (IPG), 9.rear collisionwarning (RCW), 10.image-based stopping and go (ISG), 11.automatic following navigation (AFN), and12.drowsiness detection (DD). Partial programs had been transformed to companies and partial systems hadbecome commercial products.In the above ADAS studies, we encountered several tricky problems, such as the bad detection rate offront vehicle detection in the bad weather conditions, the unstable pedestrian detection, the unreliableobstacle detection in the rear monitor, the unpredicted results in the automatic following navigation, etc. Thus,we propose a three-year research project, intending using the techniques of CNN to improve the detection andrecognition rates of our ADAS studies. In this project, we have a technique development and an ADASapplication in each year. In the first year, we will: i.develop a small CNN structure to match the ability of theIC design in Taiwan, ii.solve the problems of underfitting and overfitting, and iii.apply to the FCW system. Inthe second year, we want to: i.develop a CNN system for high-speed object detection without scanning thewhole image, and ii.apply to the pedestrian detection in the PCW system and the obstacle detection in theRCW system. In the third year, we intend to i.develop a small-sample CNN system for target detection andtracking, such that the CNN system can be easily transformed and applied among different situations, andii.apply to the guider detection and tracking in the AFN system. Moreover, we will also implement theproposed CNN systems on the mobile embedded parallel devices.This study is based on our fruitful previous results, focus on the fixed topics to develop special CNNsystems to solve the tricky problems on visual detection and recognition. The principal investigator of thisproject is an original researcher on computer vision; he has studied computer vision techniques more thanthirty years; moreover, he has the experience of computer vision applied on ADAS more than fifteen years.He has gotten 10 US, Taiwan, or China patents in these few years. Partial techniques are practiced and havebeen employed by several companies. In 2015, we collaborated with TSMC (Taiwan SemiconductorManufacturing Company) to study the CNN technique on the defect detection of scanning electronmicroscope (SEM) images. In 2016, we also studied CNN technique on human face recognition with nationalsecurity unit; thus we have ability to complete the execution of this research project.