Fire-YOLO: A Small Target Object Detection Method for Fire Inspection

Abstract

For the detection of small targets, fire-like and smoke-like targets in forest fire images, as well as fire detection under different natural lights, an improved Fire-YOLO deep learning algorithm is proposed. The Fire-YOLO detection model expands the feature extraction network from three dimensions, which enhances feature propagation of fire small targets identification, improves network performance, and reduces model parameters. Furthermore, through the promotion of the feature pyramid, the top-performing prediction box is obtained. Fire-YOLO attains excellent results compared to state-of-the-art object detection networks, notably in the detection of small targets of fire and smoke. Overall, the Fire-YOLO detection model can effectively deal with the inspection of small fire targets, as well as fire-like and smoke-like objects. When the input image size is 416 × 416 resolution, the average detection time is 0.04 s per frame, which can provide real-time forest fire detection. Moreover, the algorithm proposed in this paper can also be applied to small target detection under other complicated situations.

Ship Fire Detection Based on an Improved YOLO Algorithm with a Lightweight Convolutional Neural Network Model

“…Jiao et al [ 29 ] proposed a forest fire detection model based on improved YOLOv3. Zhao, Lei et al [ 30 ] proposed fire-YOLO model to detect small targets based on YOLO. Gagliardi et al [ 31 ] proposed a faster region-based convolutional neural network (R-CNN) to detect suspicious fire regions (SRoF) and non-fire regions based on their spatial features.…”

Handbook Science of Technology and Innovation

“…Computer vision systems have benefited from deep neural network models such as YOLO (You Only Look Once) (Redmon et al, 2016), this pre-trained model was implemented for numerous applications that require identification of objects that are contained in digital images or video. One of the applications that exists for identifying forest fires through the YOLO network is found in the work of Zhao (2022), in which developed a dataset (collection of elements) of 370 images containing fire and smoke situations with dimensions of 1850 x 1850 pixels. The authors made a modification on the convolutional network and named it Fire-YOLO, to later compare it against YOLO-V3 and Faster R-CNN (Girshick, 2015).…”

“…On the other hand, the other challenge that exists is the early detection of a fire, as well as the prediction of its evolution to avoid a rapid spread. Some points that was analyzed from the information collected previously in the works of González-Gutiérrez et al (2019), Pompa-García & González (2011), Lawrence & Zhang (2019), Hernandez-Hostaller, ( 2017) Zhao (2022) and Parajuli et al (2020) is that forest fire detection systems should consider the following elements.…”

An Enhanced Light Object Detection for Indiscernible Object in the Special Scene

“…The familiarized one-stage object detection algorithm includes OverFeat, SSD, RetinaNet, YOLOX , YOLOv1-v7 (Junos et al,2021), and so on. In the two-stage (Zhao et al,2022) method, a small set of candidate objects are pipelined work on the predicted proposal boxes to classify and identify the target object by the algorithm. There are the familiarized two-stage (Tk et al,2020) algorithms, such as R-CNN (Wu et al,2020), SPP-Net, Faster R-CNN (Avola et al,2021) and R-FCN.…”