zxiaomzxm

According to [Auto-Encoding Variational Bayes](https://arxiv.org/abs/1312.6114) `kl_terms[t]=0.5*tf.reduce_sum(mu2+sigma2-2*logsigma,1)-T*.5` should be `0.5*tf.reduce_sum(mu2+sigma2-2*logsigma-1,1)` equals to `0.5*tf.reduce_sum(mu2+sigma2-2*logsigma,1)-z_size*.5`

What is difference with version 2 and version 3 algorithm in erasor.cpp?

1

HI, nice work! I read the whole code and I found there are two version `compare_vois*` methods in the `erasor.cpp` file. So my qustion is what is difference between them?

Nice work! I'm curious about the **Architecture of Update Operator** in the ablation part, can you give me the 3-layer conv snippet you use in the paper?

Are there some results about the inference time of different model in your framework?

Hi! I have reimplement the reprojected distance loss in my own framework, but I found it hard to converge, because my initial pose is close to identity, thus in this...

https://github.com/TRI-ML/packnet-sfm/blob/6e3161f60e7161115813574557761edaffb1b6d1/packnet_sfm/losses/supervised_loss.py#L50-L53 `diff` and `diff2` seem have different shapes，thus can not be concatenated.

Question about the derivation of the CV/CA model

1

https://github.com/erikbohnsack/pmbm/blob/9fc4dbbb81898f1a9bfaa19c7e387b0cab40be28/utils/motion_models.py#L96-L99 https://github.com/erikbohnsack/pmbm/blob/9fc4dbbb81898f1a9bfaa19c7e387b0cab40be28/utils/motion_models.py#L214-L219 Hi, I'm confused about the form of the covariancce matrix `Q` in the CV/CA model, how to derivate it?