favorite31 Sobhan Naderi Parizi, Andrea Vedaldi, Andrew Zisserman, and Pedro F.
favorite28Table 1: Cat Head Detection: Results comparing DPM and GDPM with different number of positive and negative parts.
favorite14For instance, in the deformable part models (DPMs) latent variables indicate the location of the object parts .
favorite4The objective of the proposed framework is to learn a model which maximizes the evidence (characterized by positive latent variables) for the foreground class and at the same time minimizes the counter evidence (characterized by negative latent variables) lying in background class.
favorite0Finally, we experiment on a generalized form of Deformable Part Models with negative latent variables and show significant improvements on two different detection tasks..