Abstract
In the wake of developments in remote sensing, the application of target detection of remote sensing is of increasing interest. Unfortunately, unlike natural image processing, remote sensing image processing involves dealing with large variations in object size, which poses a great challenge to researchers. Although traditional multi-scale detection networks have been successful in solving problems with such large variations, they still have certain limitations: (1) The traditional multi-scale detection methods note the scale of features but ignore the correlation between feature levels. Each feature map is represented by a single layer of the backbone network, and the extracted features are not comprehensive enough. For example, the SSD network uses the features extracted from the backbone network at different scales directly for detection, resulting in the loss of a large amount of contextual information. (2) These methods combine with inherent backbone classification networks to perform detection tasks. RetinaNet is just a combination of the ResNet-101 classification network and FPN network to perform the detection tasks; however, there are differences in object classification and detection tasks. To address these issues, a cross-scale feature fusion pyramid network (CF2PN) is proposed. First and foremost, a cross-scale fusion module (CSFM) is introduced to extract sufficiently comprehensive semantic information from features for performing multi-scale fusion. Moreover, a feature pyramid for target detection utilizing thinning U-shaped modules (TUMs) performs the multi-level fusion of the features. Eventually, a focal loss in the prediction section is used to control the large number of negative samples generated during the feature fusion process. The new architecture of the network proposed in this paper is verified by DIOR and RSOD dataset. The experimental results show that the performance of this method is improved by 2–12% in the DIOR dataset and RSOD dataset compared with the current SOTA target detection methods.
Highlights
With regard to development in technology and the advent of the era of machine learning, deep learning technology is advancing by leaps and bounds and has encouraged the development of target detection technology.Traditional target detection [1,2] extracts features from candidate regions within the image using techniques such as Haar [3], HOG [4] or sparse representation [5,6,7,8] and classifies them using the SVM [9] model
The experimental results show that the performance of this method is improved by 2–12% in the DIOR dataset and RSOD dataset compared with the current SOTA target detection methods
There are two categories of deep learning-based target detection methods: the first category involves two-stage target detection based on region proposals whereas the second category involves single-stage target detection based on regression
Summary
Traditional target detection [1,2] extracts features from candidate regions within the image using techniques such as Haar [3], HOG [4] or sparse representation [5,6,7,8] and classifies them using the SVM [9] model. Deep learning-based target detection methods have been used widely. There are two categories of deep learning-based target detection methods: the first category involves two-stage target detection based on region proposals whereas the second category involves single-stage target detection based on regression. RPN instead of selective search algorithm improves detection speed. Converts the target detection task into a regression problem, greatly speeding up detection. Selective search and detection are divided into two stages resulting in slow speed; poor detection for small targets The introduction of the new feature pyramid solves the defect that the feature map of each scale in the traditional feature pyramid contains only single level or few levels of features.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
