Search

Robust segmentation in adverse weather conditions is crucial for autonomous driving. However, these scenes struggle with recognition and make annotations expensive, resulting in poor performance. As a result, the Segment Anything Model (SAM) was recently proposed to finely segment the spatial structure of scenes and to provide powerful prior spatial information, thus showing great promise in resolving these problems. However, SAM cannot be applied directly for different geographic scales and non-semantic outputs. To address these issues, we propose SAM-EDA, which integrates SAM into an unsupervised domain adaptation mean-teacher segmentation framework. In this method, we use a "teacher-assistant" model to provide semantic pseudo-labels, which will fill in the holes in the fine spatial structure given by SAM and generate pseudo-labels close to the ground truth, which then guide the student model for learning. Here, the "teacher-assistant" model helps to distill knowledge. During testing, only the student model is used, thus greatly improving efficiency. We tested SAM-EDA on mainstream segmentation benchmarks in adverse weather conditions and obtained a more-robust segmentation model. [ABSTRACT FROM AUTHOR]

The accurate detection of relevant vehicles, pedestrians, and other targets on the road plays a crucial role in ensuring the safety of autonomous driving. In recent years, object detectors based on Transformers or CNNs have achieved excellent performance in the fully supervised paradigm. However, when the trained model is directly applied to unfamiliar scenes where the training data and testing data have different distributions statistically, the model's performance may decrease dramatically. To address this issue, unsupervised domain adaptive object detection methods have been proposed. However, these methods often exhibit decreasing performance when the gap between the source and target domains increases. Previous works mainly focused on utilizing the style gap to reduce the domain gap while ignoring the content gap. To tackle this challenge, we introduce a novel method called IDI-SCD that effectively addresses both the style and content gaps simultaneously. Firstly, the domain gap is reduced by disentangling it into the style gap and content gap, generating corresponding intermediate domains in the meanwhile. Secondly, during training, we focus on one single domain gap at a time to achieve inter-domain invariance. That is, the content gap is tackled while maintaining the style gap, and vice versa. In addition, the style-invariant loss is used to narrow down the style gap, and the mean teacher self-training framework is used to narrow down the content gap. Finally, we introduce a multiscale fusion strategy to enhance the quality of pseudo-labels, which mainly focus on enhancing the detection performance for extreme-scale objects (very large or very small objects). We conduct extensive experiments on four mainstream datasets of in-vehicle images. The experimental results demonstrate the effectiveness of our method and its superiority over most of the existing methods. [ABSTRACT FROM AUTHOR]

With the increasing use of the strategy and group target attack method in the modern battlefield, multi-target inverse synthetic aperture radar (ISAR) imaging simultaneously with high efficiency draws more and more attention, which gives a promising prospect for aerospace target detection and recognition in the multi-target scenario. To overcome the shortcomings of traditional multi-target imaging with one beam at one pulse repetition time (PRT) based on phase array radar (PAR), this paper proposes a novel multi-target imaging waveform design method based on the newly full digital array radar (DAR). Firstly, we propose using radar waveform diversity with 2D orthogonality to realize multi-target ISAR imaging with high imaging quality and efficiency. Then, to meet the constant modulus requirement for maximizing the transmitting power, orthogonal frequency division multiplexing (OFDM) chirp theory is proposed to directly generate the transmit waveform instead of the traditional optimization method with the nonconvex problem for waveform design. Based on time-variant weighted and time diversity technology, a of group transmit waveforms is designed, which can form multiple beams simultaneously and make the signals arriving at different targets approximately orthogonal. Finally, simulations and experiments are carried out to demonstrate the effectiveness of the proposed method. [ABSTRACT FROM AUTHOR]

The use of electronically steered antennas in the azimuth dimension typically leads to a staircase-like antenna beam steering law in the Terrain Observation by Progressive Scan (TOPS) wide-swath synthetic aperture radar (SAR) data acquisition mode, which will introduce paired echoes in the focused images. This paper proposes a new approach for removing such paired echoes from TOPS SAR images based on the generalization of the ideal optimum filtering concept, which can be implemented easily in the SAR data processing. Modeling the amplitude-modulated azimuth signal shows that the absolute phase of the introduced paired echoes cannot be determined due to the random rotation angle jump time for each target, which will prevent the precise use of optimum filtering. An extended optimum filtering approach, which is originally proposed for suppressing the azimuth ambiguities in SAR images, is reintroduced in this particular case, and a new approximated and generalized form of the deconvolving filtering in the approach is redefined to accommodate the undetermined phase for both the strongest paired distortion peaks and the other peripheral peaks in the distorted impulse response function (IRF). Simulated data from a TOPS SAR mode with staircase-like beam steering are used to verify the improvement in image quality by using the new method. [ABSTRACT FROM AUTHOR]