Plot heatmap for viewpoints using KDE(kernel density estimation), so that we can know the distribution of poses.
Plot heatmap for viewpoints using KDE(kernel density estimation), so that we can know the distribution of poses.
To visualize viewpoints distribution in 6D pose dataset, I first convert rotation matrix to viewpoint in polar coordinate system.Let ρ be the distance of the optical center to the origin, θ and φ be the longitude and latitude, respectively.I use kernel density estimation (KDE) to estimate the non-parametric distributions of ρ,θ,φ for each category respectively from the LINEMOD Occluded dataset.
The the above graph is generated by estimate_viewpoint.py.