Traditional real-time object detection networks deployed in autonomous aerial vehicles (AAVs) struggle to extract features from small objects in complex backgrounds with occlusions and overlapping objects. To address ...
详细信息
Traditional real-time object detection networks deployed in autonomous aerial vehicles (AAVs) struggle to extract features from small objects in complex backgrounds with occlusions and overlapping objects. To address this challenge, we propose FM-RTDETR, a real-time object detection algorithm optimized for small object detection. We redesign the encoder of RT-DETRv2 by integrating the Feature Aggregation and Diffusion Network (FADN), improving the algorithm's ability to capture contextual information. Subsequently, we introduce the Parallel Atrous Mamba Feature Fusion Module (PAMFFM), which combines shallow and deep semantic information to better capture small object features. Furthermore, we propose the Cross-stage Enhanced Feature Fusion Module (CEFFM), merging features for small objects to provide richer and more detailed information. Finally, we propose STIoU Loss, which incorporates a penalty term to adjust the scaling of the loss function, improving detection granularity for small objects. FM-RTDETR achieves AP(50) scores of 54.0% and 56.3% on the VisDrone2019-DET and AI-TOD datasets. Compared with other state-of-the-art methods, our method shows great potential in small object detection from drones.
An M-ary time reversal (TR) maximum likelihood classifier for a single pair of transmitting and receiving transducer element was derived in [1] for underwater acoustic target detection applications. This paper conside...
详细信息
ISBN:
(纸本)9781424426768
An M-ary time reversal (TR) maximum likelihood classifier for a single pair of transmitting and receiving transducer element was derived in [1] for underwater acoustic target detection applications. This paper considers a more general TR setup consisting of a P-element transmitting array and an N element receiving array and derives the M-ary conventional and TR classifiers for the multielement case in an electromagnetic communication environment. We show that the TR algorithm provides a classification gain of over 3 dB at low signal to noise ratios as compared to the conventional classifiers.
This paper deals with the joint signal and parameter estimation for linear state-space models. An efficient solution to this problem can be obtained by using a recursive instrumental variable technique based on two du...
详细信息
This paper deals with the joint signal and parameter estimation for linear state-space models. An efficient solution to this problem can be obtained by using a recursive instrumental variable technique based on two dual Kalman filters. In that case, the driving process and the observation noise in the state-space representation for each filter must be white with known variances. These conditions, however, are too strong to be always satisfied in real cases. To relax them, we propose a new approach based on two dual H infin filters. Once a new observation of the disturbed signal is available, the first H infin algorithm uses the latest estimated parameters to estimate the signal, while the second H infin algorithm uses the estimated signal to update the parameters. In addition, as the H infin filter behavior depends on the choice of various weights, we present a way to recursively tune them. This approach is then studied in the following cases: (1) consistent estimation of the AR parameters from noisy observations and (2) speech enhancement, where no a priori model of the additive noise is required for the proposed approach. In each case, a comparative study with existing methods is carried out to analyze the relevance of our solution.
Compressed Sensing (CS) has been used in ECG signal compressing with the rapid development of real-time & dynamic ECG applications. signal reconstruction process is an essential step in CS-based ECG processing. Ma...
详细信息
ISBN:
(纸本)9781509008964
Compressed Sensing (CS) has been used in ECG signal compressing with the rapid development of real-time & dynamic ECG applications. signal reconstruction process is an essential step in CS-based ECG processing. Many recovery algorithms have been reported in the last decades. However, the comparative study on their reconstructing performances for CS-based ECG signalprocessing lacks, especially in real-time applications. This study aimed to investigate this issue and provide useful information. Four typical recovery algorithms, i.e., compressed sampling matching pursuit (CoSaMP), orthogonal matching pursuit (OMP), expectation-maximum-based block sparse Bayesian learning (BSBLEM) and bound-optimization-based block sparse Bayesian learning (BSBL BO) were compared. Two performance indices, i.e., the percentage of root-mean-square difference (PRD) and the reconstructing time (RT), were tested to observe their changes with the change of compression ratio (CR). The results showed that BSBL_BO and BSBL_EM methods had better performances than OMP and CoSaMP methods. More specifically, BSBL_BO reported the best PRD results while BSBL_EM achieved the best RT index.
Operations an digital signals similar to those employed for Z transforms are used for producing parallel multidimensional algorithms. All operations are given in the time domain and are part of the Unified signal Alge...
详细信息
Operations an digital signals similar to those employed for Z transforms are used for producing parallel multidimensional algorithms. All operations are given in the time domain and are part of the Unified signal Algebra. algorithms are given using dataflow type diagrams and are presented for the parallel specification of convolution, triple convolution, Volterra convolution, and correlation.
Two major tasks of partial discharges (PD) measurements may be distinguished, (i) providing general evidence and the type of PD (detection) and (ii) the location of the PD. Dependent on the type of device under test t...
详细信息
ISBN:
(纸本)9781424416219
Two major tasks of partial discharges (PD) measurements may be distinguished, (i) providing general evidence and the type of PD (detection) and (ii) the location of the PD. Dependent on the type of device under test the two issues have changing priority. For the on-line/on-site PD location in power transformers unconventional PD measuring methods like acoustic ultra-sonic measurements or electromagnetic measurements up to the UHF (ultra high frequencies) range are performed, while the Time Domain Reflectometry (TDR) of electric PD signals is a standard technique for the location of PD in power cables.
We introduce holographic (optically inspired) algorithms, which are suited for implementation on massively parallel, locally interconnected arrays of nanoscale devices. This computing method is inspired by optical sig...
详细信息
We introduce holographic (optically inspired) algorithms, which are suited for implementation on massively parallel, locally interconnected arrays of nanoscale devices. This computing method is inspired by optical signalprocessing, but it neither relies on optical wave propagation nor optical hardware. We describe implementations on digital semiconductor circuitry and on magnetoelectronic devices.
A new concept for solving statistical synthesis problems of radiometric devices and systems (RDS) is developed. The reasonability of using ultra-wideband (UWB) RDSs is substantiated. The basic data of mathematical app...
详细信息
A new concept for solving statistical synthesis problems of radiometric devices and systems (RDS) is developed. The reasonability of using ultra-wideband (UWB) RDSs is substantiated. The basic data of mathematical apparatus of V-transforms employed when processing space-time UWB fields and their statistical characteristics is given. A wide class of UWB RDS statistical synthesis problems (compensation, zero-type, modified modulation type, modulation compensation type, multi-antenna systems including cross-correlation-compensation systems and aperture synthesis systems) of stationary and scanning types are solved. The appropriateness of using the Kravchenko weighting functions for refinement of primary radiometric images in radiometric scanning devices is substantiated.
暂无评论