The acquisition of large-scale paired visible and thermal images is crucial for enhancing face recognition systems, especially in low-light environments where visible spectrum images fail. However, the task is hindered by the scarcity of thermal images and the need for identity consistency during image generation. In this paper, we propose Pair-ID, an innovative framework that addresses these challenges by creating a shared latent space for simultaneous generation of paired visible and thermal images. Pair-ID integrates identity information into text embeddings and employs fixed templates for diverse facial poses, streamlining the customization process and reducing computational demands. The framework's Joint Learner encodes both modalities, facilitating synchronized image generation and preserving facial details. Extensive evaluations show that Pair-ID surpasses current methods in efficiency and performance for paired data generation, making it a promising solution for face recognition under varying lighting conditions.
14 March 2024 in IEEE Transactions on Instrumentation and Measurement
Process monitoring (PM) plays an important role in industrial intelligent maintenance. However, existing PM methods usually assume that the noise satisfies Gaussian or non-Gaussian, making it difficult to deal with monitoring data contaminated by noise with unknown statistics in nonlinear dynamic processes. If these methods are used directly, there will be a certain number of undetectable areas. To address this issue, we propose a new end-to-end monitoring framework by integrating a recurrent neural network (RNN) with complementary dual residual generators, which to the best of our knowledge has not been investigated in PM. Technically, the pre residual generator targets to capture the potential dynamics and nonlinear relationships but may generate non-Gaussian residuals, while the post residual generator targets to complement the pre residual generator by training with a negentropy loss function. It is guaranteed that although the noise statistics are unknown, the generated outputs can be combined into a residual satisfying the Gaussian distribution. Hence, we can easily monitor processes using the T 2 test statistics. Extensive numerical examples with unknown noise statistics, including the simulated Tennessee Eastman (TE) process and a real-world marine diesel engine (DE) process, demonstrate the Gaussianity of generated residuals and verify the superiority compared to state-of-the-art methods.
There is little known about kinetic changes in anterior cruciate ligament deficiency (ACLD) combined with meniscal tears during jogging. Therefore, 29 male patients with injured ACLs and 15 healthy male volunteers were recruited for this study to investigate kinetic abnormalities in male patients after ACL deficiency combined with a meniscal injury during jogging. Based on experimental data measured by an optical tracking system, a subject-specific musculoskeletal model was employed to estimate the tibiofemoral joint kinetics during jogging. Between-limb and interpatient differences were compared by the analysis of variance. The results showed that decreased knee joint forces and moments of both legs in ACLD patients were detected during the stance phase compared to the control group. Meanwhile, compared with ACLD knees, significantly fewer contact forces and flexion moments in ACLD combined with lateral and medial meniscal injury groups were found at the mid-stance, and ACLD with medial meniscal injury group showed a lower axial moment in the loading response (p < 0.05). In conclusion, ACLD knees exhibit reduced tibiofemoral joint forces and moments during jogging when compared with control knees. A combination of meniscus injuries in the ACLD-affected side exhibited abnormal kinetic alterations at the loading response and mid-stance phase.
14 May 2026 in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Synthetic Aperture Radar (SAR) ship detection plays a key role in the field of remote sensing. Data augmentation is frequently used to enhance detector performance due to the common challenge of limited annotated data. However, most augmented methods are limited to geometric transformations, “copy-paste”, or image inpainting, which are guided by strict constraints and suffer limited target and background diversity. To address this, we propose a novel multi-phase transfer-learned framework with layout-conditioned diffusion model. It incorporates sparse target layouts as relaxed conditions into the latent diffusion model, where the conditions take the form of bounding boxes that specify target positions and sizes. This leaves other ship attributes (e.g. orientation, scattering characteristics) and backgrounds to be freely generated and enhance target and background diversity. For enhancing the model's convergence on small datasets, we first pre-train it on large-scale datasets and then fine-tune it on small-scale ones for transfer learning. This enables precise layout-guided target generation even when using small datasets. For subsequent data augmentation, we adopt the widely used grounded-language-to-image generation (GLIGEN) model for practical use, and based on the transfer-learned model, we further develop a new automatic augmentation pipeline with a filtering mechanism. This pipeline integrates multiple foundation models and tools, including layout augmentation, segment anything model (SAM)-based annotation correction, and data filtering, and can automatically adjust inappropriate annotations and screen high-quality data to serve as augmented data for downstream detection tasks. Experiments on SSDD and HRSID datasets validate that the pipeline improves multiple detectors' performance across in-shore and off-shore scenarios, and qualitative analyses confirm high diversity and quality of generated images.
Spiking Neural Networks (SNNs) as one of the biology-inspired models have received much attention recently. It can significantly reduce energy consumption since they quantize the real-valued membrane potentials to 0/1 spikes to transmit information thus the multiplications of activations and weights can be replaced by additions when implemented on hardware. However, this quantization mechanism will inevitably introduce quantization error, thus causing catastrophic information loss. To address the quantization error problem, we propose a regularizing membrane potential loss (RMP-Loss) to adjust the distribution which is directly related to quantization error to a range close to the spikes. Our method is extremely simple to implement and straightforward to train an SNN. Furthermore, it is shown to consistently outperform previous state-of-the-art methods over different network architectures and datasets.