Shenzhen University

Sijia Lai

Qiang Li

Lei Huang

Xiaopeng Li

Peichang Zhang

Weize Sun

Lifang Feng

Enfu Ye

Provided are a lidar three-dimensional (3D) target detection method based on region pillar feature fusion and a system thereof. The method includes: collecting a point cloud data set of a target region and a coordinate thereof, determining a point cloud range and dividing the point cloud range into several sub-regions; calculating a maximum difference of a z coordinate in each sub-region, creating a feature array, and allocating the maximum difference to points in the sub-region to form a pillar feature. According to an area of a target, the allocation number is dynamically adjusted to obtain the target feature. The detection performance of the network for small targets is improved through region attention enhancement feature coding.

Lidar 3D target detection method based on region pillar feature fusion and system thereof

Growing community of inventors

Shenzhen, China

Sijia Lai