Motion pattern-based crowd scene classification using histogram of angular deviations of trajectories

Automated crowd behaviour analysis and monitoring is a challenging task due to the unpredictable nature of the crowd within a particular scene and across different scenes. The prior knowledge of the type of scene under consideration is a crucial mid-level information, which could be utilized to develop robust crowd behaviour analysis systems. In this paper, we propose an approach to automatically detect the type of a crowded scene based on the global motion patterns of the objects within the scene. Three different types of scenes whose global motion pattern characteristics vary from uniform to non-uniform are considered in this work, namely structured, semi-structured, and unstructured scenes, respectively. To capture the global motion pattern characteristics of an input crowd scene, we first extract the motion information in the form of trajectories using a key-point tracker and then compute the average angular orientation feature of each trajectory. This paper utilizes these angular features to introduce a novel feature vector, termed as Histogram of Angular Deviations (HAD), which depicts the distribution of the pair-wise angular deviation values for each trajectory vector. Since angular deviation information is resistant to changes in scene perspectives, we consider it as a key feature for distinguishing the scene types. To evaluate the effectiveness of the proposed HAD-based feature vector in classifying the crowded scenes, we build a crowd scene classification model by training the classical machine learning algorithms on the publicly available Collective Motion Database. The experimental results demonstrate the superior crowd classification performance of the proposed approach as compared to the existing methods. In addition to this, we propose a technique based on quantizing the angular deviation values to reduce the feature dimension and subsequently introduce a novel crowd scene structuredness index to quantify the structuredness of an input crowded scene based on its HAD.

Multi-View Urban Scene Classification with a Complementary-Information Learning Model

Traditional urban scene-classification approaches focus on images taken either by satellite or in aerial view. Although single-view images are able to achieve satisfactory results for scene classification in most situations, the complementary information provided by other image views is needed to further improve performance. Therefore, we present a complementary information-learning model (CILM) to perform multi-view scene classification of aerial and ground-level images. Specifically, the proposed CILM takes aerial and ground-level image pairs as input to learn view-specific features for later fusion to integrate the complementary information. To train CILM, a unified loss consisting of cross entropy and contrastive losses is exploited to force the network to be more robust. Once CILM is trained, the features of each view are extracted via the two proposed feature-extraction scenarios and then fused to train the support vector machine classifier for classification. The experimental results on two publicly available benchmark data sets demonstrate that CILM achieves remarkable performance, indicating that it is an effective model for learning complementary information and thus improving urban scene classification.

A Lightweight Convolutional Neural Network Based on Group-Wise Hybrid Attention for Remote Sensing Scene Classification

With the development of computer vision, attention mechanisms have been widely studied. Although the introduction of an attention module into a network model can help to improve e classification performance on remote sensing scene images, the direct introduction of an attention module can increase the number of model parameters and amount of calculation, resulting in slower model operations. To solve this problem, we carried out the following work. First, a channel attention module and spatial attention module were constructed. The input features were enhanced through channel attention and spatial attention separately, and the features recalibrated by the attention modules were fused to obtain the features with hybrid attention. Then, to reduce the increase in parameters caused by the attention module, a group-wise hybrid attention module was constructed. The group-wise hybrid attention module divided the input features into four groups along the channel dimension, then used the hybrid attention mechanism to enhance the features in the channel and spatial dimensions for each group, then fused the features of the four groups along the channel dimension. Through the use of the group-wise hybrid attention module, the number of parameters and computational burden of the network were greatly reduced, and the running time of the network was shortened. Finally, a lightweight convolutional neural network was constructed based on the group-wise hybrid attention (LCNN-GWHA) for remote sensing scene image classification. Experiments on four open and challenging remote sensing scene datasets demonstrated that the proposed method has great advantages, in terms of classification accuracy, even with a very low number of parameters.