Best overall performance was obtained by denoising method D1 (Hough transformation) followed by modelling method M2 (IRTLS cylinder fit), as this combination produced the lowest RMSE and bias values, keeping a high stem segment detection rate (Fig. 8).
Clarify what the example dbh_algo (https://github.com/tiagodc/TreeLS/blob/9033693503ee71ace2335a6b8fa2dce864df37cd/README.md) is in the paper (https://www.sciencedirect.com/science/article/pii/S0168169917301114?via%3Dihub)? How to realize the combinations in the paper?
Is it this?