CVAM-Pose: Conditional Variational Autoencoder for Multi-Object Monocular Pose Estimation