Pedestrian Detection in Crowded Environments through Bayesian Prediction of Sequential Probability Matrices

Abstract

In order to safely navigate populated environments, an autonomous vehicle must be able to detect human shapes using its sensory systems, so that it can properly avoid a collision. In this paper, we introduce a Bayesian approach to the Viola-Jones algorithm, as a method to automatically detect pedestrians in image sequences. We present a probabilistic interpretation of the basic execution of the original tool and develop a technique to produce approximate convolutions of probability matrices with multiple local maxima.

Vision-Based Human Detection Techniques: A Descriptive Review

“…Ultimately, all these key points are finding blobs. The size of the blob is determined by the value of σ. Viola-Jones object detection, face detection [53], pedestrian detection [57] • takes less computational time while maintaining high accuracy in real-time [53] • can detect object under complex situation (i.e. rain , snow) [53] • can successfully detect pedestrian in low resolution [54] • lacks the entire image fine details [55], [56] • texture or shape information are ignored [55], [56] • sensitive to lighting condition [55], [56] • unsuitable for general object detection [55], [56] SIFT object recognition, face recognition [58], gesture recognition [59], video tracking [60], motion tracking [61] • robust to occlusion, clutter and noise [62] • distinctive features [62] • performance is close to real-time [63] • flexible to extend with other features [64] • poor performance with lighting changes and blur [65] • computationally expensive [65] PCA-SIFT object recognition, image retrieval [66], image analysis [65] • reduces the dimensionality of the SIFT descriptors [66] • improves the matching accuracy and speed in real-world environment [66] • sensitive to viewpoint change [65] • color information is ignored [67] SURF object recognition [51], [68], face detection [69], image registration [66], object classification [70] • takes less time for computation and feature matching [51] • improves the robustness of feature extraction [51] • struggl...…”

“…They are parallel to each other and shown in Figure 11. Based on these conditions the pair of hyperplanes can be found at maximum margin by minimizing ||u||2, related to constraints (57). However, the result of two dimensional (2D) cases is described in Figure (11a…”

“…There are some constraints (57); to solve this problem a Lagrangian formulation is employed where positive Lagrange multipliers β i , i = 1, , l, constraints of the form d i ≥ 0; are introduced. To form the Lagrangian, the constraint equations are multiplied by positive Lagrange multipliers and subtracted from the objective function.…”

Integarted Minimum Cost Sub-Block Matching Distance based Face Recognition using Internet of Things

“…So there is a need to control partial occlusion in order to achieve substantial performance in face recognition. The literature [2][3][4][5][6][7][8][9][10][11][12][13][14][15] focus on finding occlusion-tolerant features or classifiers to reduce the effect of partial occlusions in face recognition. Though, the hidden features from the occluded parts can still deteriorate the recognition performance.…”

Improving the Lane Reference Detection for Autonomous Road Vehicle Control

“…The three most common methods for detecting longrange obstacles are radar [5][6][7][8], laser scanner [9][10][11][12][13], and computer vision [14,15]. Each of them has advantages and disadvantages [16] and sensor fusion is commonly used to overcome the individual limitations [17][18][19][20][21].…”