Hi author, thank you for the impressive work! While exploring the code, I was looking for the BEV-VAE model architecture but only found simple-densenet, which appears to be a very simple fully-connected network. It doesn’t seem to match the multi-view inputs and sophisticated feature-fusion process described in the paper. Could you please clarify whether the complete BEV-VAE implementation has not yet been released?
Hi author, thank you for the impressive work! While exploring the code, I was looking for the BEV-VAE model architecture but only found simple-densenet, which appears to be a very simple fully-connected network. It doesn’t seem to match the multi-view inputs and sophisticated feature-fusion process described in the paper. Could you please clarify whether the complete BEV-VAE implementation has not yet been released?