Fire detection is an important task in many applications. Smoke and flame are two essential symbols of fire in images. In this paper, we propose an algorithm to detect smoke and flame simultaneously for color dynamic video sequences obtained from a stationary camera in open space. Motion is a common feature of smoke and flame and usually has been used at the beginning for extraction from a current frame of candidate areas. The adaptive background subtraction has been utilized at a stage of moving detection. In addition, the optical flow-based movement estimation has been applied to identify a chaotic motion. With the spatial and temporal wavelet analysis, Weber contrast analysis and color segmentation, we achieved moving blobs classification. Real video surveillance sequences from publicly available datasets have been used for smoke detection with the utilization of our algorithm. We also have conducted a set of experiments. Experiments results have shown that our algorithm can achieve higher detection rate of 87% for smoke and 92% for flame.

In this paper, a person tracking algorithm for indoor video surveillance is presented. The algorithm contains the following steps: person detection, person features formation, features similarity calculation for the detected objects, postprocessing, person indexing, and person visibility determination in the current frame. Convolutional Neural Network (CNN) YOLO v3 is used for person detection. Person features are formed based on H channel in HSV color space histograms and a modified CNN ResNet. The proposed architecture includes 29 convolutional and one fully connected layer. As the output, it forms a 128-feature vector for every input image. CNN model was trained to perform feature extraction. Experiments were conducted using MOT methodology on stable camera videos in indoor environment. Main characteristics of the presented algorithm are calculated and discussed, confirming its effectiveness in comparison with the current approaches for person tracking in an indoor environment. Our algorithm performs real time processing for object detection and tracking using CUDA technology and a graphics card NVIDIA GTX 1060.

For practical use, the relevance of indoor surveillance from multiple cameras to track the movement of people and reidentify them in video sequences is constantly increasing. This is a complex task due to the effect of uneven illumination, background inhomogeneity, overlap, uncertainty of the trajectories of people, and the similarity of their visual features. The article presents an approach to track people by video sequences and reidentify them in multicamera video surveillance systems that are used indoors. At the first step, people are detected using a YOLO v4 convolution neural network (CNN) and described by a rectangular area. Further, the search for the face area and the calculation of its features are carried out, which in the developed method are used when accompanying a person in a video sequence and during his intercamera reidentification. This approach improves the accuracy of tracking with a complex movement trajectory and multiple intersections of people with similar characteristics. The search for faces is carried out on the detected areas based on the multitasking MTCNN, and the MobileFaceNetwork model is used to form the vector of the features of the face. Human features are generated using a modified CNN based on ResNet34 and an HSV color tone channel histogram. The correspondence between people on different frames is established based on the analysis of the spatial coordinates of faces and people, as well as their CNN features, using the Hungarian algorithm. To ensure the accuracy of intercamera tracking, reidentification is performed based on the facial features. Five test video sequences of different numbers of people captured indoors with a fixed video camera were used to test and compare different approaches. The obtained experimental results confirmed the strength of the characteristics of the proposed approach.

In this paper, we improve the accuracy of person re-identification in images obtained from distributed video surveillance systems by choosing activation functions for convolutional neural networks. The most popular activation functions used for object detection, namely, ReLU, Leaky-ReLU, PReLU, RReLU, ELU, SELU, GELU, Swish, and Mish, are analyzed based on the following metrics: Rank1, Rank5, Rank10, mAP, and training time. For feature extraction, ResNet-50, DenseNet-121, and DarkNet-53 architectures are employed. The experimental study is carried out on open datasets Market1501 and PolReID. The accuracy of person re-identification is assessed after thrice-repeated training and testing with different activation functions, neural network architectures, and datasets by averaging the values of the selected metrics.

Some similarity functions for comparing the features of objects in the processing of static images and video sequences are proposed. These functions provide the possibility to find the normalized similarity value and are determined by calculating the ratios between the minimum and maximum values for all the pairs of analyzed features. To find the complex value characterizing the similarity of compared images as a whole, the summation or multiplication of calculated ratios is used. It is proposed to take into account the distances between features for such types of calculations. Some results of experimental studies on the comparison of the qualitative characteristics of similarity functions, their robustness against different types and levels of noises, and the possibility of the precise localization of objects on an image for the case when the brightness levels of pixels are used as features are presented.