Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
Technische Universität Braunschweig
3
0
4
Expensive inference-time diffusion is no longer necessary for zero-shot OOD detection: using a foundation segmenter as a label-prompted spatial verifier sets a new state of the art at a fraction of the compute.
StaticSegFormer boosts segmentation frame rates by 34% without compromising accuracy, challenging the dominance of dynamic pruning methods.
HadamardNet not only enhances adversarial robustness in object detection and segmentation but also achieves state-of-the-art disturbance detection in a single pass.