3D Single Object Tracking
22 papers with code • 0 benchmarks • 1 datasets
3D tracking of a single object, based on an initial 3D bounding box, provided to the tracker. 3D single object tracking is commonly performed using point cloud data from Lidars, as it provides valuable depth information, which is lost in camera images. However, irregular point cloud structure and an increasing point sparsity with distance makes Lidar-based 3D single object tracking a nontrivial task.
Benchmarks
These leaderboards are used to track progress in 3D Single Object Tracking
Libraries
Use these libraries to find 3D Single Object Tracking models and implementations
KITTI
Most implemented papers
Multi-Correlation Siamese Transformer Network with Dense Connection for 3D Single Object Tracking
Instead of performing correlation of the two branches at just one point in the network, in this paper, we present a multi-correlation Siamese Transformer network that has multiple stages and carries out feature correlation at the end of each stage based on sparse pillars.
SeqTrack3D: Exploring Sequence Information for Robust 3D Point Cloud Tracking
Most existing methods perform tracking between two consecutive frames while ignoring the motion patterns of the target over a series of frames, which would cause performance degradation in the scenes with sparse points.
