Video flame detection algorithm based On multi-feature fusion technique

Pedestrians, motorist, and cyclist remain the victims of poor vision and negligence of human drivers, especially in the night. Millions of people die or sustain physical injury yearly as a result of traffic accidents. Detection and recognition of road markings play a vital role in many applications such as traffic surveillance and autonomous driving. In this study, we have trained a nighttime road-marking detection model using NIR camera images. We have modified the VGG-16 base network of the state-of-the-art faster R-CNN algorithm by using a multilayer feature fusion technique. We have demonstrated another promising feature fusion technique of concatenating all the convolutional layers within a stage to extract image features. The modification boosts the overall detection performance of the model by utilizing the advantages of the shallow layers and the deep layers of the VGG-16 network. The training samples were augmented using random rotation and translation to enhance the heterogeneity of the detection algorithm. We have achieved a mean average precision (mAP) of 89.48% and 92.83% for the baseline faster R-CNN and our modified method, respectively.

Download Full-text

Video Flame Detection Algorithm Based on Improved GMM and Multi-Feature Fusion

Laser & Optoelectronics Progress ◽

10.3788/lop202158.0410006 ◽

2021 ◽

Vol 58 (4) ◽

pp. 0410006

Author(s):

张驰 Zhang Chi ◽

孟庆浩 Meng Qinghao ◽

井涛 Jing Tao

Keyword(s):

Feature Fusion ◽

Detection Algorithm ◽

Flame Detection

Download Full-text

Multi-feature fusion based fast video flame detection

Building and Environment ◽

10.1016/j.buildenv.2009.10.017 ◽

2010 ◽

Vol 45 (5) ◽

pp. 1113-1122 ◽

Cited By ~ 81

Author(s):

Juan Chen ◽

Yaping He ◽

Jian Wang

Keyword(s):

Feature Fusion ◽

Flame Detection

Download Full-text

Pedestrian detection algorithm based on improved muti-scale feature fusion

Journal of Physics Conference Series ◽

10.1088/1742-6596/2078/1/012008 ◽

2021 ◽

Vol 2078 (1) ◽

pp. 012008

Author(s):

Hui Liu ◽

Keyang Cheng

Keyword(s):

Clustering Algorithm ◽

Feature Fusion ◽

Pedestrian Detection ◽

Detection Algorithm ◽

Data Sets ◽

False Detection ◽

Scale Feature ◽

Multi Scale ◽

Dilated Convolution ◽

Small Targets

Abstract Aiming at the problem of false detection and missed detection of small targets and occluded targets in the process of pedestrian detection, a pedestrian detection algorithm based on improved multi-scale feature fusion is proposed. First, for the YOLOv4 multi-scale feature fusion module PANet, which does not consider the interaction relationship between scales, PANet is improved to reduce the semantic gap between scales, and the attention mechanism is introduced to learn the importance of different layers to strengthen feature fusion; then, dilated convolution is introduced. Dilated convolution reduces the problem of information loss during the downsampling process; finally, the K-means clustering algorithm is used to redesign the anchor box and modify the loss function to detect a single category. The experimental results show that the improved pedestrian detection algorithm in the INRIA and WiderPerson data sets under different congestion conditions, the AP reaches 96.83% and 59.67%, respectively. Compared with the pedestrian detection results of the YOLOv4 model, the algorithm improves by 2.41% and 1.03%, respectively. The problem of false detection and missed detection of small targets and occlusion has been significantly improved.

Download Full-text

Multimodal plant recognition through hybrid feature fusion technique using imaging and non-imaging hyper-spectral data

Journal of King Saud University - Computer and Information Sciences ◽

10.1016/j.jksuci.2018.09.018 ◽

2018 ◽

Cited By ~ 3

Author(s):

Pradip Salve ◽

Pravin Yannawar ◽

Milind Sardesai

Keyword(s):

Spectral Data ◽

Feature Fusion ◽

Fusion Technique ◽

Hyper Spectral ◽

Plant Recognition

Download Full-text

Financial Topic Detection Algorithm Based on Multi-feature Fusion

Lecture Notes in Electrical Engineering - Proceedings of 2020 Chinese Intelligent Systems Conference ◽

10.1007/978-981-15-8458-9_2 ◽

2020 ◽

pp. 11-19

Author(s):

Qiang Zhang ◽

Junping Du ◽

Feifei Kou

Keyword(s):

Feature Fusion ◽

Detection Algorithm ◽

Topic Detection

Download Full-text

Detection and Tracking Cows by Computer Vision and Image Classification Methods

International Journal of Security and Privacy in Pervasive Computing ◽

10.4018/ijsppc.2021010101 ◽

2021 ◽

Vol 13 (1) ◽

pp. 1-45

Author(s):

Terry Gao

Keyword(s):

Feature Fusion ◽

Binary Classification ◽

Classification Problem ◽

Detection Algorithm ◽

Video Sequences ◽

Image Block ◽

Body Contour ◽

Adaboost Algorithm ◽

Target Model ◽

Detection And Tracking

In this paper, the cow recognition and traction in video sequences is studied. In the recognition phase, this paper does some discussion and analysis which aim at different classification algorithms and feature extraction algorithms, and cow's detection is transformed into a binary classification problem. The detection method extracts cow's features using a method of multiple feature fusion. These features include edge characters which reflects the cow body contour, grey value, and spatial position relationship. In addition, the algorithm detects the cow body through the classifier which is trained by Gentle Adaboost algorithm. Experiments show that the method has good detection performance when the target has deformation or the contrast between target and background is low. Compared with the general target detection algorithm, this method reduces the miss rate and the detection precision is improved. Detection rate can reach 97.3%. In traction phase, the popular compressive tracking (CT) algorithm is proposed. The learning rate is changed through adaptively calculating the pap distance of image block. Moreover, the update for target model is stopped to avoid introducing error and noise when the classification response values are negative. The experiment results show that the improved tracking algorithm can effectively solve the target model update by mistaken when there are large covers or the attitude is changed frequently. For the detection and tracking of cow body, a detection and tracking framework for the image of cow is built and the detector is combined with the tracking framework. The algorithm test for some video sequences under the complex environment indicates the detection algorithm based on improved compressed perception shows good tracking effect in the changing and complicated background.

Download Full-text