The detection of small objects by oriented bounding box in aerial images is a recent hot topic. However, since the aerial images are not collected at the same height, the Ground Sample Distance (GSD) is different for each image, so that small objects are easily overlooked. Existing algorithms are designed for multi-scale object detection, and feature fusion is time-consuming, resulting in a large amount of model parameters that is not easy to deploy on embedded devices. We propose three methods to address the above problems. First, we scale the collected aerial images to the same scale according to the GSD value. Second, we change the structure of Feature Pyramid Network (FPN) and only keep the necessary low-level feature maps. Finally, we rescale the anchor for the specific scene. We validate our proposed method on the DOTA dataset. The results show that the modified model using our method can identify more small-scale objects, and the maximum number of model parameters can be reduced by 2.7%, the inference speed can be increased by 13.24%, and the model size was reduced by up to 28% when the detection accuracy is the same as the original algorithm.
An artificial intelligent decision-making system based on Deep Q Network is developed according to the characteristic of the optoelectronic countermeasures for defense. In view of the high complexity of the input state variables of the system, simulation method is used to sift the state variables so as to reduce the input dimension of the network. In addition, simulation method is used to generate enough samples for the network training. Aiming at the adaptability evaluation of the system, evolutionary evaluation index is designed and simulation method is used to evaluate the adaptability of online learning ability of the system.
For the linear array scanning infrared detection system, the reasonable design of the system hardware architecture and data processing flow is the key to ensure the system to achieve real-time target detection and fast recognition. Fast and effective target recognition algorithm is the core of the system design. The signal processing of the linear scanning infrared detection system designed in this paper adopts the hardware architecture of FPGA + DSP + GPU, and puts forward the false target discrimination method of sky and earth line based on semantic segmentation based on deep learning, which is different from the traditional threshold detection and segmentation method based on artificial template matching. The deep learning method uses the semantic information and spatial information of infrared image and has certain adaptability. Finally, the algorithm is implemented on the hardware system through the field measured data, and the effectiveness of the algorithm is verified.
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.