![]() However, the generalization capacity of expert-driven learning approaches is often very limited, and some of them are designed to solve specific problems. Many efforts have been made to develop shape features that are not sensitive to brightness, orientation, or scale. The algorithm of pattern extraction is designed or adjusted by experts, according to the discipline of image processing or the features of the object itself. ![]() Feature extraction of the human body in the image is carefully designed in order to transform the raw image into a feature map, which can be recognized by the classifiers. In the expert-driven learning approaches, feature extraction and classification are two necessary steps. Ĭurrent research on human detection can be broadly classified into two categories: Expert-driven and data-driven learning approaches. Besides, the performance of the detection is also affected by the high temperature of the cluttered background caused by hot weather in the day time and radiation from other sources like the sun or the lamps. First, the performance of the far-infrared pedestrian detection could be influenced by low image resolution, low contrast, and the large noises of far-infrared images. However, there are two main factors that have hindered the development of detection based on the far-infrared imagery. Among these bands, humans are more physically visible in the far-infrared camera than in the other cameras. In general, the IR spectrum can be classified into four sub-bands, such as near-IR, short-wave IR, medium-wave IR, far-infrared. Although an increasing number of theories and methods have been put forward as solutions for visible light classification and detection problems, those for IR imagery detection have never been proposed in a systematical manner. Human candidate pixels are supposed to be with higher intensity in the IR images because the temperature of the human body is usually higher than that of the background. In the infrared (IR) spectrum approach, the IR cameras receive radiation emitted from the scene in the daytime and nighttime conditions, and their intensity distribution is not sensitive to the illumination of the scene or color of the object. ![]() Therefore, such types of detection can be used not only in intelligent surveillance systems, but also in the advanced driver assistance systems (ADAS) for greater safety for humans and vehicles. ![]() Moreover, information from infrared imagery can be employed both day and night. The infrared radiation of some wave band can penetrate through the cloud and mist. This information, especially those from thermal radiation, is available to be employed to detect targets with higher temperatures. Of these, the specialized use of infrared imagery for pedestrian detection in images is of particular interest, since they contain different information from the visible light images. Different types of sensing devices can be used to capture information from diverse dimensions. Pedestrian detection is a vital research topic in the field of computer vision, one of significant theoretical interest, and with various applications. The results prove that background suppression and suitable feature expansion will accelerate the training process and enhance the performance of IR image-based deep learning models. The experimental results show that the Mean Average Precisions (mAPs) of four different datasets have been increased by 5.22% on average. Four different experiments are performed from various perspectives in order to gauge the efficiency of our approach. A precise fusion algorithm is designed to combine the information from different visual saliency maps in order to reduce the effect of truncation and miss detection. Besides, a novel background suppression method is proposed to stimulate the attention principle of human vision and shrink the region of detection. In this paper, a novel channel expansion technique based on feature fusion is proposed to enhance the IR imagery and accelerate the training process. However, two problems in deep learning-based detection are the implicit performance and time-consuming training. Recent studies on pedestrian detection in infrared (IR) imagery have employed data-driven approaches. Pedestrian detection is an important task in many intelligent systems, particularly driver assistance systems.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |