Ultra-large-scale offshore photovoltaic (PV) installations require efficient and reliable construction-phase inspection to ensure installation integrity and compliance with engineering specifications. As the deployment scale expands to thousands of platforms and millions of photovoltaic modules, conventional manual inspection becomes labor-intensive, time-consuming, and increasingly prone to omission errors. This study presents an autonomous inspection framework based on AI-driven computer vision for the detection and localization of missing photovoltaic modules in offshore PV systems. The proposed framework integrates high-resolution UAV-acquired RGB imagery, YOLOv8-based object detection, geographic coordinate transformation, spatial deduplication, and deterministic grid-based indexing to convert aerial observations into structured engineering inspection records. Each detected missing module is automatically assigned a unique platform identifier together with row–column coordinates, enabling engineering-level localization while eliminating redundant detections caused by overlapping UAV imagery. The proposed framework was validated using a dataset comprising 2800 annotated UAV images collected from a 1 GW offshore photovoltaic project. The experimental results revealed a recall of 96.15%, an F1-score of 98.04%, and a manual verification consistency of 96.83%. Geographic deduplication eliminated duplicate grid records, while the average processing time of 1.12 s per image demonstrates the computational feasibility of the framework for large-scale offshore deployment. The results confirm that integrating deep learning-based visual detection with geographic spatial mapping enables reliable, scalable, and engineering-oriented verification of missing photovoltaic modules during construction-phase inspection, thereby supporting standardized and data-driven acceptance workflows for large-scale renewable energy infrastructure.

The installation of grouted connections in deep-water offshore wind turbine jackets faces severe visibility constraints and limited diver operation time, necessitating an unmanned, real-time monitoring solution using remotely operated vehicles (ROVs). However, the lack of dedicated datasets and the challenge of balancing segmentation accuracy with edge-deployment efficiency remain major obstacles. This paper addresses these gaps by constructing, to the best of our knowledge, the first underwater grout overflow instance segmentation dataset reported in the published literature, comprising 2088 images and 5733 polygon instances annotated under a two-level semantic system (“overflow” and “overflow warning”). On this basis, we propose a collaborative optimization method based on YOLOv11n-seg and validate it with a strict 12-fold leave-one-group-out cross-validation (LOGOCV; 2088 images, video-level disjoint groups, unweighted macro averages). We first design a lightweight Multi-Path Feature Extractor (MPFE), MPFE-lite, retaining only the B3 asymmetric strip-convolution branch (depthwise 1 × 5 and 5 × 1 kernels). MPFE-lite improves mask mAP@0.5 over the YOLOv11n-seg baseline by 3.85 percentage points (35.9% → 39.8%; paired t-test p = 0.017, Wilcoxon p = 0.027, 10 wins out of 12 folds) while using 11.6% fewer parameters (2.54 M vs. 2.88 M) and slightly fewer GFLOPs (9.80 vs. 9.90). The full three-branch MPFE is also significantly better than baseline (mask mAP@0.5 +3.53 points, p = 0.007/0.002), but adding its dilated multi-scale and fixed Sobel branches to B3 gives no significant cross-group gain; the Dual-Branch Channel-Spatial Attention (DCSA) attention module and DySample upsampling, which appeared beneficial on a single fixed split, likewise produce no significant LOGOCV improvement and are therefore removed. For deployment-oriented compression, a Bidirectional Feature Pyramid Network (BiFPN)-based variant (1.21 M parameters, 7.38 GFLOPs) offers a 58.0% parameter reduction at a cost of 3.10 mask-mAP points, providing an optional lightweight configuration. Against peer lightweight models over an eight-fold partial comparison, MPFE-lite is significantly more accurate than YOLOv5n-seg and YOLOv6s-seg and comparable to YOLOv8n-seg while being lighter. Gradient-weighted Class Activation Mapping (Grad-CAM) visualizations provide qualitative supporting evidence consistent with the strip-convolution design rather than definitive confirmation of its mechanism. This work provides a complete engineering closed-loop from model inference to on-site visualization, filling critical data and algorithmic gaps in underwater grout monitoring.