Confidence alone is often misleading in hyperspectral image classification, as models tend to mistake high predictive scores for correctness while lacking awareness of uncertainty. This leads to confirmation bias, especially under sparse annotations or class imbalance, where models overfit confident errors and fail to generalize. We propose CABIN (Cognitive-Aware Behavior-Informed learNing), a semi-supervised framework that addresses this limitation through a closed-loop learning process of perception, action, and correction. CABIN first develops perceptual awareness by estimating epistemic uncertainty, identifying ambiguous regions where errors are likely to occur. It then acts by adopting an Uncertainty-Guided Dual Sampling Strategy, selecting uncertain samples for exploration while anchoring confident ones as stable pseudo-labels to reduce bias. To correct noisy supervision, CABIN introduces a Fine-Grained Dynamic Assignment Strategy that categorizes pseudo-labeled data into reliable, ambiguous, and noisy subsets, applying tailored losses to enhance generalization. Experimental results show that a wide range of state-of-the-art methods benefit from the integration of CABIN, with improved labeling efficiency and performance.

The task of infrared small target detection holds significant application value in military surveillance and civilian security. Small targets typically exhibit characteristics such as weak texture, low contrast, and small scale, which make them susceptible to interference from complex backgrounds. Although existing deep learning-based models have incorporated background modeling to enhance contextual awareness, high-frequency and low-frequency information are often entangled during feature propagation. The entanglement makes it difficult to distinguish targets from the background, consequently leading to false alarms and missed detections. To address the issue, we propose a dual-band detection network based on spectral decoupling, termed the Frequency-Guided Semantic Understanding Network (FSUNet) for infrared small target detection. The encoder and decoder of our network are composed of Wavelet-Guided Semantic Disentangling Blocks (W-SD) and Dual-Band Spectral Cooperative Blocks (D-SC), respectively. The W-SD explicitly separates high-frequency edge and low-frequency structural semantic features using wavelet frequency-domain priors. The D-SC is implemented by combining a Dual-Band Refinement (DBR) and a Semantic Re-Sampling (SRS) mechanism. The DBR filters and optimizes effective information for the high-frequency and low-frequency branches separately, suppressing noise interference. The SRS fuses features from the two branches, by dynamically correcting their statistical differences and reconstructing channel weights to highlight discriminative features, thereby enhancing target distinguishability and noise-resistant robustness. Experimental results on public datasets SIRST, NUDT-SIRST, and IRSTD-1K demonstrate that our FSUNet method outperforms existing methods from the perspective of detection accuracy and robustness, especially in complex backgrounds with low contrast. The code can be available on https://github.com/fulongcai/FSUNet-main.

Infrared small target detection (ISTD) is a challenging task due to the low contrast and small size of the targets, which are often affected by complex backgrounds. UNet and its variants, known for their encoder–decoder structures, are widely used in such tasks since they can capture both local and global features. However, a significant drawback of UNet-based networks is the irreversible loss of crucial local features during downsampling, leading to missed detections and false positives, especially for small targets. Compared to other architectures like feature pyramid networks, UNet provides a more symmetric and efficient structure, allowing it to handle dense pixel-wise predictions effectively. However, standard UNet models still struggle to fully retain small target details, motivating the need for further improvements. To address this issue, we propose HintU, a novel network to recover the local features lost by various UNet-based methods for effective ISTD. HintU has two key contributions. First, it introduces the “Hint” mechanism for the first time, i.e., leveraging the prior knowledge of target locations to highlight critical local features. Second, it improves the mainstream UNet-based architecture to preserve target pixels even after downsampling. HintU can shift the focus of various networks (e.g., vanilla UNet, UNet++, UIUNet, MiM+, and HCFNet) from the irrelevant background pixels to a more restricted area from the beginning. Experimental results on three datasets NUDT-SIRST, SIRSTv2, and IRSTD1K demonstrate that HintU enhances the performance of existing methods with only an additional 1.88-ms cost (on RTX Titan). Additionally, the explicit constraints of HintU enhance the generalization ability of UNet-based methods. Code is available athttps://github.com/Wuzhou-Quan/HintU.

To reduce computational redundancies, a common approach is to integrate detection and re-identification (Re-ID) into a single network in multi-object tracking (MOT), referred to as “tracking by detection.” Most of the previous research has focused on resolving the conflict between the detection and Re-ID branches, considering it a simple coupling. In our work, we uncover that the entangled state between the detection and Re-ID tasks is much more complex than previous idea, resulting in a form of competition that degrades performance. To address the preceding issue, we propose a feature disentanglement network that deeply disentangles the intricately interwoven latent space of features and provides differentiated feature maps for each individual task. Furthermore, considering the demand for shallow semantic features in the feature re-ID branch, we also introduce a feature re-globalization module to enrich the shallow semantics. By integrating two distinct networks into a one-shot online MOT method, we develop a robust MOT tracker (named HDGTrack ). We conduct extensive experiments on a number of benchmarks, and our experimental results demonstrate that our method significantly outperforms state-of-the-art MOT methods. Besides, HDGTrack is efficient and can run at 13.9 (MOT17) and 8.7 (MOT20) frames per second.

Pansharpening plays a crucial role in the domain of remote sensing image processing, as it allows for the generation of high-resolution multispectral images. The main objective of pansharpening is to fuse low-resolution multispectral and high-resolution panchromatic images, resulting in high-resolution multispectral images that exhibit uniform spectra and enhanced spatial details. Consequently, the primary focus of related research is to preserve accurate features from both input images and achieve superior image reconstruction. In this paper, we introduce a novel pansharpening framework called Dual Texture-Edge Maintaining Transformer (DTEMPan). Our framework achieves exceptional fusion results by leveraging a novel, more interpretable, and powerful architecture that considers pansharpening as dual, distinct deep sub-semantic branches. It independently reconstructs sub-semantic layer information, leading to improved performance. The DTEMPan architecture incorporates a dual transformer design comprising shared perception encoders and two parallel, effective semantic-level decoders. The hybrid multi-scale texture maintaining decoder and the precise edge maintaining decoder are responsible for reconstructing the general low-frequency and rare high-frequency signals, respectively. Through the integration of complementary information from both decoders, DTEMPan is capable of reconstructing accurate edges and high spatial information while preserving rich spectral details. Extensive experimental evaluations have demonstrated it significantly outperforms state-of-the-art methods on a number of benchmarks. Our code is available at https://github.com/D-Walter/DTEMPan.