Early Smoke Detection Based on Improved YOLO-PCA Network

Abstract

Early detection of smoke having indistinguishable pixel intensities in digital images is a difficult task. To better maintain fire surveillance, early smoke detection is crucial. To solve the problem, we have integrated the principal component analysis (PCA) as a pre-processing module with the improved version of You Only Look Once (YOLOv3). The ordinary YOLOv3 structure has been improved after inserting one extra detection scale at stride-4 specifically to detect immense small smoke instances in the wild. The improved network design establishes a sequential relation between feature maps of lower spatial information and fine-grained semantic information in up-sampled maps via skip connections and concatenation operations. The testing of the improved model is carried out on self-prepared smoke datasets. In digital images, the smoke instances are captured in various complicated environments, for example, the mountains and fog in the background. A principal component analysis (PCA) helps in useful features selection and abandons the involvement of redundant features in the testing of the trained network hence, overcoming the latency at inference stage. In addition, to process small smoke images as positive samples during training, new sizes of anchors are calculated on small smoke data at a specified Intersection over Union (IoU) threshold. The experimental results show the improvement in precision rate, recall rate, and mean harmonic (F1-score) by 2.67, 3.06, and 5.59 percentages. The respective improvements in average precision (AP) and mean average precision (mAP) are 1.66 and 2.78 percentages.

AI-Enhanced Gas Flares Remote Sensing and Visual Inspection: Trends and Challenges

“…One typical realization of aforementioned methods are Convolutional Neural Networks (CNNs)-based structure, and Two-dimensional (2D) CNNs are predominantly employed for the purpose of detecting smoke in images, whereas Three-Dimensional (3D) CNNs are primarily utilized for smoke detection in videos [39]- [41]. More recently, You Only Look Once (YOLO)-based [42], [43] structures are applied in this smoke detection field and YOLO now is at its 8th version [44]. Another subgroup for smoke detection is transformerbased structure such as Detection Transformer [45], [46], and temporal event transformers [47], [48].…”

“…The nascent smoke is hard to be discovered both because of its light color, bad contrast with the background, and the distance [49]. Besides, since there is no benchmark (as different groups own different private data sets and can not build up each other), the instance number of existing early smoke data sets is limited [42] [50] [51], this fact makes the research even more difficult.…”

Automatic detection of bulldozer-induced changes on a sandy beach from video using YOLO algorithm

“…The novelty of the presented study relies in three main points: Change images, obtained using PCA are used as the input for YOLO detection. Although PCA has been used to reduce dimensionality prior to YOLO detection (Masoom et al, 2022), to the knowledge of the authors there are no previous studies that combine PCA change detection and YOLO to identify changing objects in video images.…”

The Development of Symbolic Expressions for Fire Detection with Symbolic Classifier Using Sensor Fusion Data

“…The smoke particles in the air are disrupting the current, which activates the alarm. A literature investigation showed that the majority of scientific papers available online regarding fire detection from smoke detection involve the implementation of various deep convolutional neural networks (DCNN) such as ResNet [ 4 ], Inception [ 5 , 6 ], YOLO [ 7 ], and others. In these papers, the image/video data types were used.…”