The ability to classify urban objects in large urban scenes from point clouds efficiently and accurately still remains a challenging task today. A new methodology for the effective and accurate classification of terrestrial laser scanning (TLS) point clouds is presented in this paper. First, in order to efficiently obtain the complementary characteristics of each 3-D point, a set of point-based descriptors for recognizing urban point clouds is constructed. This includes the 3-D geometry captured using the spin-image descriptor computed on three different scales, the mean RGB colors of the point in the camera images, the LAB values of that mean RGB, and the normal at each 3-D point. The initial 3-D labeling of the categories in urban environments is generated by utilizing a linear support vector machine classifier on the descriptors. These initial classification results are then first globally optimized by the multilabel graph-cut approach. These results are further refined automatically by a local optimization approach based upon the object-oriented decision tree that uses weak priors among urban categories which significantly improves the final classification accuracy. The proposed method has been validated on three urban TLS point clouds, and the experimental results demonstrate that it outperforms the state-of-the-art method in classification accuracy for buildings, trees, pedestrians, and cars.

The quantity and quality of cropland are the key to ensuring the sustainable development of national agriculture. Remote sensing technology can accurately and timely detect the surface information, and objectively reflect the state and changes of the ground objects. Using high-resolution remote sensing images to accurately extract cropland is the basic task of precision agriculture. The traditional model of cropland semantic segmentation based on the deep learning network is to down-sample high-resolution feature maps to low resolution, and then restore from low-resolution feature maps to high-resolution ideas; that is, obtain low-resolution feature maps through a network, and then recover to high resolution by up-sampling or deconvolution. This will bring about the loss of features, and the segmented image will be more fragmented, without very clear and smooth boundaries. A new methodology for the effective and accurate semantic segmentation cropland of high spatial resolution remote sensing images is presented in this paper. First, a multi-temporal sub-meter cropland sample dataset is automatically constructed based on the prior result data. Then, a fully convolutional neural network combined with contextual feature representation (HRNet-CFR) is improved to complete the extraction of cropland. Finally, the initial semantic segmentation results are optimized by the morphological post-processing approach, and the broken spots are ablated to obtain the internal homogeneous cropland. The proposed method has been validated on the Jilin-1 data and Gaofen Image Dataset (GID) public datasets, and the experimental results demonstrate that it outperforms the state-of-the-art method in cropland extraction accuracy. We selected the comparison of Deeplabv3+ and UPerNet methods in GID. The overall accuracy of our approach is 92.03%, which is 3.4% higher than Deeplabv3+ and 5.12% higher than UperNet.

To reduce the cost of manually annotating training data for parsing outdoor scenes, we propose a supervised approach with automatically generating training data for classifying 3-D point clouds of large-scale urban scenes. In this approach, the input point cloud is aggregated into point clusters, and the disjoint set union issue is combined with geometric attributes of each point cluster to obtain object segments. The prior knowledge among different classes is used to label the segments by using the decision-tree model. Then, the initialized training samples are generated automatically. The confidence estimation for the labeling is employed to filter the mislabeled training samples. With the generated training data, we train a random forest classifier to create the initial classification of the 3-D scene on the set of descriptors for each 3-D point. The classification results are further optimized by multilabel conditional Random Fields. Experimental results on five urban point clouds captured by different types of scanners (i.e., terrestrial laser scanning, vehicle laser scanning, and airborne laser scanning datasets) demonstrate that the proposed approach achieves a competitive classification performance.

Effectively leveraging multi-source unmanned aerial vehicle (UAV) observations for reliable object recognition is often compromised by environmental extremes (e.g., occlusion and low illumination) and the inherent physical discrepancies between modalities. To overcome these limitations, we propose DLANet, a lightweight, rotation-aware multimodal object detection framework that introduces a dual-level attention relearning strategy to maximize complementary information from visible (RGB) and thermal infrared (TIR) imagery. DLANet integrates two novel components: the Implicit Fine-Grained Fusion Module (IF²M), which facilitates deep cross-modal interaction by jointly modeling channel and spatial dependencies at intermediate stages, and the Adaptive Branch Feature Weighting (ABFW) module, which dynamically recalibrates modality contributions at higher levels to suppress noise and pseudo-targets. This synergistic approach allows the network to relearn feature importance based on real-time scene conditions. To support industrial applications, we construct the OilLeak dataset, a dedicated benchmark for onshore oil-spill detection. The experimental results demonstrate that DLANet achieves state-of-the-art performance, recording an mAP_0.5 of 0.858 on the public DroneVehicle dataset while maintaining high efficiency, with 39.04 M parameters and 72.69 GFLOPs, making it suitable for real-time edge deployment.