Accurate detection of citrus can be easily affected by adjacent branches and overlapped fruits in natural orchard conditions, where some specific information of citrus might be lost due to the resultant complex occlusion. Traditional deep learning models might result in lower detection accuracy and detection speed when facing occluded targets. To solve this problem, an improved deep learning algorithm based on YOLOv5, named IYOLOv5, was proposed for accurate detection of citrus fruits. An innovative Res-CSPDarknet network was firstly employed to both enhance feature extraction performance and minimize feature loss within the backbone network, which aims to reduce the miss detection rate. Subsequently, the BiFPN module was adopted as the new neck net to enhance the function for extracting deep semantic features. A coordinate attention mechanism module was then introduced into the network’s detection layer. The performance of the proposed model was evaluated on a home-made citrus dataset containing 2000 optical images. The results show that the proposed IYOLOv5 achieved the highest mean average precision (93.5%) and F1-score (95.6%), compared to the traditional deep learning models including Faster R-CNN, CenterNet, YOLOv3, YOLOv5, and YOLOv7. In particular, the proposed IYOLOv5 obtained a decrease of missed detection rate (at least 13.1%) on the specific task of detecting heavily occluded citrus, compared to other models. Therefore, the proposed method could be potentially used as part of the vision system of a picking robot to identify the citrus fruits accurately.

Effective underwater sensing is crucial for environmental protection and sustainable energy transitions, particularly as we face growing challenges in marine ecosystem monitoring, resource management, and the need for efficient energy infrastructure. To support these efforts, we propose a multimodal sensing approach that enhances underwater detection and distance estimation by combining affordable sonar technology with stereo vision-based depth cameras. Our method integrates the Ping 360 single-beam sonar for target detection and distance measurement with depth refinement from the Intel RealSense D455 camera. A promptable segmentation model automates sonar target detection, overcoming challenges such as acoustic noise and shadowing without requiring large labeled datasets. Depth images from the stereo camera are enhanced using a Depth-Anything model, addressing underwater-specific issues like noise, missing regions, and light attenuation, achieving accurate depth maps for distances up to 1.2 meters underwater. By leveraging multimodal sensing, this approach not only improves underwater robotics for navigation, manipulation, and exploration but also plays a key role in monitoring and maintaining energy infrastructure, such as offshore wind farms and underwater pipelines. Accurate, real-time sensing of these installations ensures more efficient operations, minimizes the environmental impact, and aids in the sustainable management of ocean resources. This enables better energy production and resource utilization, which are essential for a smarter and more sustainable energy transition.