3 accepted papers
The F-measure is a widely used metric in multi-label classification, where multiple labels are predicted simultaneously for a single instance. The optimal prediction rule for F-measure requires estimating q^2+1 probabilities, where q is the number of labels. Existing approaches train q multinomial e
Semantic segmentation labels each pixel in an image with its corresponding class, and is typically evaluated using the Intersection over Union (IoU) and Dice metrics to quantify the overlap between predicted and ground-truth segmentation masks. In the literature, most existing methods estimate pixel…
Vehicle-to-everything (V2X) communication techniques enable the collaboration between vehicles and many other entities in the neighboring environment, which could fundamentally improve the perception system for autonomous driving. However, the lack of a public dataset significantly restricts the res