Search papers, labs, and topics across Lattice.
School of Computer Science and Technology, Soochow University
4
0
4
Low-light crowd counting accuracy is revolutionized by a novel multi-modal approach that leverages depth and edge information, outperforming existing methods.
RT-Counter achieves 7.4x faster performance in object counting while maintaining competitive accuracy, breaking the traditional accuracy-speed trade-off.
Real-world conditions can severely impair object counting accuracy, but a novel test-time training approach boosts performance without requiring architectural changes.
MambaCount achieves state-of-the-art object counting accuracy with linear complexity, challenging the dominance of Transformer-based methods in dense visual environments.
