Cascaded Cross-Modality Fusion Network for 3D Object Detection

Chen, Zhiyu; Lin, Qiong; Sun, Jing; Feng, Yujian; Liu, Shangdong; Liu, Qiang; Ji, Yimu; Xu, He

doi:10.3390/s20247243

Cited by 2 publications

(1 citation statement)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The resulting digital terrain models and digital surface models depict the surrounding reality in detail. These features determine the multidirectional use of airborne laser scanning in various fields of science, such as engineering solutions-calculation 3D displacements of bridges [3], 3D object detection along the road [4,5], building extraction [6,7], land cover change detection, and forest succession monitoring [8,9] for heterogeneous land use urban mapping [10], coastal monitoring [11,12], or archeological research [13,14].…”

Section: Literature Reviewmentioning

confidence: 99%

Using LiDAR System as a Data Source for Agricultural Land Boundaries

Borowiec

Marmol

2022

Remote Sensing

View full text Add to dashboard Cite

In this study, LiDAR sensor data were used to identify agricultural land boundaries. This is a remote sensing method using a pulsating laser directed toward the ground. This study focuses on accurately determining the edges of parcels using only the point cloud, which is an original approach because the point cloud is a scattered set, which may complicate finding those points that define the course of a straight line defining the parcel boundary. The innovation of the approach is the fact that no data from other sources are supported. At the same time, a unique contribution of the research is the attempt to automate the complex process of detecting the edges of parcels. The first step was to classify the data, using intensity, and define land use boundaries. Two approaches were decided, for two test fields. The first test field was a rectangular shaped parcel of land. In this approach, pixels describing each edge of the plot separately were automatically grouped into four parts. The edge description was determined using principal component analysis. The second test area was the inner subdivision plot. Here, the Hough Transform was used to emerge the edges. Obtained boundaries, both for the first and the second test area, were compared with the boundaries from the Polish land registry database. Performed analyses show that proposed algorithms can define the correct course of land use boundaries. Analyses were conducted for the purpose of control in the system of direct payments for agriculture (Integrated Administration Control System—IACS). The aim of the control is to establish the borders and areas of croplands and to verify the declared group of crops on a given cadastral parcel. The proposed algorithm—based solely on free LiDAR data—allowed the detection of inconsistencies in farmers’ declarations. These mainly concerned areas of field roads that were misclassified by farmers as subsidized land, when in fact they should be excluded from subsidies. This is visible in both test areas with areas belonging to field roads with an average width of 1.26 and 3.01 m for test area no. 1 and 1.31, 1.15, 1.88, and 2.36 m for test area no. 2 were wrongly classified as subsidized by farmers.

show abstract

Section: Literature Reviewmentioning

confidence: 99%

Using LiDAR System as a Data Source for Agricultural Land Boundaries

Borowiec

Marmol

2022

Remote Sensing

View full text Add to dashboard Cite

show abstract

CaLiJD: Camera and LiDAR Joint Contender for 3D Object Detection

Lyu,

Qi,

You

et al. 2024

Remote Sensing

View full text Add to dashboard Cite

Three-dimensional object detection has been a key area of research in recent years because of its rich spatial information and superior performance in addressing occlusion issues. However, the performance of 3D object detection still lags significantly behind that of 2D object detection, owing to challenges such as difficulties in feature extraction and a lack of texture information. To address this issue, this study proposes a 3D object detection network, CaLiJD (Camera and Lidar Joint Contender for 3D object Detection), guided by two-dimensional detection results. CaLiJD creatively integrates advanced channel attention mechanisms with a novel bounding-box filtering method to improve detection accuracy, especially for small and occluded objects. Bounding boxes are detected by the 2D and 3D networks for the same object in the same scene as an associated pair. The detection results that satisfy the criteria are then fed into the fusion layer for training. In this study, a novel fusion network is proposed. It consists of numerous convolutions arranged in both sequential and parallel forms and includes a Grouped Channel Attention Module for extracting interactions among multi-channel information. Moreover, a novel bounding-box filtering mechanism was introduced, incorporating the normalized distance from the object to the radar as a filtering criterion within the process. Experiments were conducted using the KITTI 3D object detection benchmark. The results showed that a substantial improvement in mean Average Precision (mAP) was achieved by CaLiJD compared with the baseline single-modal 3D detection model, with an enhancement of 7.54%. Moreover, the improvement achieved by our method surpasses that of other classical fusion networks by an additional 0.82%. In particular, CaLiJD achieved mAP values of 73.04% and 59.86%, respectively, thus demonstrating state-of-the-art performance for challenging small-object detection tasks such as those involving cyclists and pedestrians.

show abstract

Cascaded Cross-Modality Fusion Network for 3D Object Detection

Cited by 2 publications

References 32 publications

Using LiDAR System as a Data Source for Agricultural Land Boundaries

Using LiDAR System as a Data Source for Agricultural Land Boundaries

CaLiJD: Camera and LiDAR Joint Contender for 3D Object Detection

Contact Info

Product

Resources

About