Pyramidal Part-Based Model for Partial Occlusion Handling in Pedestrian Classification

Abstract

Pedestrian detection and classification are of increased interest in the intelligent transportation system (ITS), and among the challenging issues, we can find limitations of tiny and occluded appearances, large variation of human pose, cluttered background, and complex environment. In fact, a partial occlusion handling is important in the case of detecting pedestrians, in order to avoid accidents between pedestrians and vehicles, since it is difficult to detect when pedestrians are suddenly crossing the road. To solve the partial occlusion problem, a pyramidal part-based model (PPM) is proposed to obtain a more accurate prediction based on the majority vote of the confidence score of the visible parts by cascading the pyramidal structure. The experimental results on the proposed scheme achieved 96.25% accuracy on the INRIA dataset and 81% accuracy on the PSU (Prince of Songkla University) dataset. Our proposed model can be applied in the real-world environment to classify the occluded part of pedestrians with the various information of part representation at each pyramid layer.

Pedestrian classification on transfer learning based deep convolutional neural network for partial occlusion handling

“…Additionally, there are a variety of applications, including video surveillance, intelligent transportation systems, and self-driving assistance systems. Though there are many significant improvements for detecting pedestrians, the crucial yet challenging problems include the large variety of poses, appearances, sizes, and types of occlusions [4]- [6]. Among these problems, partial occlusion frequently occurs due to the diversity of the partially occluded patterns of the pedestrians, and it needs further investigation between the pedestrians and the crowded instances.…”

“…Furthermore, on a large-scale ImageNet dataset, the deep CNN models were extensively tweaked for image classification and obtained great classification performance. However, the performance of the deep CNN can influence the results on a small amount of training data and the presence of partially occluded pedestrian areas in the training data [4].…”

“…Deep convolutional neural networks require hundreds of thousands of sample images, and the number of samples has an effect on learning performance. Transfer learning in deep learning is an approach whereby the training time expenses of the deep CNN model is being solved, and a small number of training samples is also solved for the specific complex models [4], [24], [27]. Furthermore, transfer learning minimizes the time required to train a neural network model and can lead to fewer generalization errors.…”

Occupancy Analysis of the Outdoor Football Fields

“…In [134], Shu et al developed a partbase model to detect partial occlusion in multi-person tracking prob-lems. Solutions to deal with partial occlusion for detection, classification and tracking applications have also been reported in [135][136][137]. In these works, the Bhattacharyya distance, pyramidal part-based model and restricted Boltzmann machine-based deep models are developed to detect occlusion.…”