Low-light image enhancement is an essential task in image restoration. Inspired by the diffusion model, the related methods have achieved remarkable results in low-level visual tasks. However, such methods are susceptible to large-scale images, generating problems such as overconsumption of resources and low recovery efficiency. To address this, we propose a detail-guided latent space low-light image enhancement diffusion model called DLDiff. Leveraging the generative power of the latent diffusion model, we explore ways to speed up inference better while producing excellent perceptual fidelity. Specifically, we initially employ the latent diffusion model to transform low-light image features into a latent space representation, thereby reducing computational resource consumption. Next, we design a lightweight detail prompt module that combines cross-convolution and vast-receptive-field convolution blocks. This module enhances the fine-grained details of the image, effectively supplements multiscale feature information, and minimizes feature loss in the latent space. Furthermore, we devise the content-aware loss group to facilitate learning noise and image information, enhancing the model's recovery capability, guiding stable sampling, and constraining diverse content generation. Through extensive experiments, we demonstrate the model's significant efficiency and quality advantages in low-light image enhancement tasks. The code is available at https://github.com/Yuii666/DLDiff .

The popularity of the Internet of Things has led to the rapid development of the consumer electronics industry and the interconnection of digital devices in various scenarios. The effective utilization of sensors can provide accurate data support for the development of IoT. With the construction of intelligent IoT, general sensors have been unable to meet the actual demand of system operation, it is necessary to rely on intelligent sensors to fully promote the intelligent and modern construction of IoT. In particular, intelligent sensors are usually employed in complex environments, which require automated monitoring of operating status to ensure the stability of electronic devices. Meanwhile, intelligent sensors are applied in industrial production, and its status monitoring is of great significance for social safety. For this purpose, we propose an automated monitoring method for intelligent sensors in consumer electronics by a joint self-attention mechanism and residual network. We design residual feature extraction and self-attention mechanism for mining salient representations of sensor faults, respectively, and fuse time-domain contextual information using long-and short-term memory network. The proposed methodology is able to capture and analyze the characteristics of sensor faults more effectively and understand the temporal nature of sensor data more comprehensively, which further improves the accuracy and reliability of operation state monitoring and fault diagnosis. The qualitative and quantitative experiments in two public datasets demonstrate the effectiveness of the proposed method.

In the 5G era and beyond, new applications (APPs) have more stringent requirements for low latency. The cloud radio access network (C-RAN) with TDM-PON-based mobile fronthaul (MFH) and multi-access edge computing (MEC) has been regarded as an efficient architecture for serving latency-critical APPs. Many studies have focused on latency optimization issues in the MFH or MEC segment. However, the ultimate latency from the end-user’s perspective is end-to-end (E2E) latency, which includes the latency in both the MFH and MEC segments. When coordinating MFH with MEC, the E2E latency optimization may be mismatched and inefficient. To address this issue, we propose an APP-aware MAC scheduling scheme that allocates uplink time resources based on the MEC-backed conditions and APPs’ traffic dynamics. Accordingly, we introduce an AI-based traffic prediction approach to capture traffic fluctuations in various APPs. Simulations show that our proposed scheme improves the performances of latency violation ratio and average E2E latency by up to 30% and 8%, respectively.

This review summarizes the research on the end effectors of agricultural harvesting robots (2010–2025) and extracts two core design principles. First of all, the selection of end effectors must follow the biological characteristics of fruits: rigid grippers are suitable for hard skinned and regular fruits; soft grippers can reduce the damage of fragile crops to a certain extent; suction cups are suitable for smooth, barrier free surfaces; the envelope type is suitable for soft and lossless picking scenes; the combined suction and grip design is more suitable for unstructured environments. Secondly, the separation mode should match the characteristics of the stem: motion separation (torsion/pull) is suitable for weak stems, while cutting is mainly used for hard stems. Unlike previous literature, this review provides a field deployability checklist (including dust/water proofing, cleanliness, maintenance, aging prevention, and aspiration prevention) to narrow the results of the laboratory and the real field environment. The three future directions of multimodal perception, variable stiffness driving and reinforcement learning are logically related to the analysis in this paper: multimodal perception optimizes the perception limit, variable stiffness solves the rigid–flexible trade-off, and reinforcement learning provides adaptive strategies for different crops. This framework can match the end effector design with the crop-specific field conditions.