Lightweight design is a key way to realize the engineering application of deep learning algorithms on Unmanned Aerial Vehicle (UAV) platform. Aiming at the low detection accuracy of the current real-time object detection algorithms of the UAV platform, a lightweight model based on Vision Transformer (ViT) is designed in this paper. Firstly, a small Convolutional Neural Network (CNN) is used to extract primary features for reducing the number of parameters and computation amount of ViT network in this model, and using window modeling to replace part of the global modeling. Then, a feature-level mask self-supervised training method is applied to pre-train the ViT structure, which helps to accelerate the convergence and avoid a lot of labeling work. Finally, the result compared with other UAV lightweight object detection algorithms in the visdrone2018 dataset shows that this model has higher average accuracy on ensuring real-time speed, and verifies the effectiveness and reference value of the lightweight design method proposed in this paper.
With the development of unmanned aerial vehicle (UAV) in aeronautical monitoring field, the performance requirements are continuously improved, each application scene also puts forward higher and higher requirements for target detection accuracy and speed. The traditional target imaging method is difficult to meet the image quality requirements, and the artificial target recognition method can’t cope with the rapid changes in the detection environment. Combined with the development of deep learning and polarization hyperspectral imaging technology, a ground target detection method based on Faster R-CNN was proposed. We proposed region proposal network (RPN) module for model training. In the target detection phase, the proposed feature map is obtained by pooling operation of interest regions. Finally, we used the proposed feature map to complete the target category classification. Three scale models were used in the experiment, and through polarization hyperspectral camera, the image data of target in different scene conditions was acquired in indoor and outdoor simulation environment for training and validation of models. The experimental results showed that the proposed method could achieve ideal detection accuracy and speed when the ground target was effectively detected.
In the target detection process of polarization optics imaging, due to the turbulent effect of the target signal transmitted in the atmosphere and the photoelectric conversion of optical imaging sensors and other factors, Salt-and-Pepper noise which affects the detection accuracy. According to the statistical characteristics of the Salt-and-Pepper noise probability density, a new structure preserved polarization image Salt-and-Pepper noise removal method is proposed. With the new signal sparse representation theory and image inpainting method, only the noise regions is restored by the noise point detecting. In the inpainting process, the structural similarity is considered which can improve the structural information retention ability of polarization image. Numerical simulation results demonstrate the validity of the proposed method both subjectively and objectively.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.