One of the nice contributions of SVO is that it has proposed a way to use 2d edge features in pose optimization by point-to-line constraint. But in the actual code, I could not find where it is assigning any feature with edge type property. I thought the feature detector part does it but it was not. The results from the feature detector do not say whether the feature is edge type or not. Anybody know how it works?
The open source version of SVO does not have the added option of the edgelets. That can found in SVO2 which is closed source. However, there is a fork of SVO where someone added the edgelet capability which can be found here.