Topics

1

General / Computing GCP marking errors offline

« on: February 15, 2022, 12:13:25 PM »

Hi,

I've built a web portal to mark GCPs and then consume this data to run automated processing on Metashape. But this works mostly like marking GCPs on Metashape before performing Aligning. I mean, I can't show average pixel error or lat/long/alt error when a user is marking GCPs. Showing these errors would help to mark the GCPs more accurately.

I understand that to show these errors, we first need to align/match the photos. Considering that aligning is already done, I want to show GCP marking errors by determining whether the points being marked on two images are same or not.

So far, I am able to use Metashape Python API and retrieve `rotation covariance`, `location covariance` and `transform` for each of the photos. Also, I am able to use Metashape API to compute pixel errors.

But I don't want to use Metashape API here as this is not feasible to be interacted from web UI. I need help in knowing if there is a way to use image transformation matrices (which can be exported from Metashape beforehand) to compute pixel error (given the marked coordinates from 3 or more images).

2

Bug Reports / Using Created (never classified) in Classify Ground Points

« on: August 04, 2021, 03:55:18 PM »

Context: I'm trying to reclassify ground points without doing Reset Classification.

My assumption: Using "Created (never classified)" as source ("From") in "Classify Ground Points" should do the same job as "Reset Classification" and selecting source as "Any Class"

Output: This setting tries to classify ground points on top of previously classified points instead of doing a fresh new classification.

Please clarify if my assumption is correct or if this is a bug.
My requirement is to do "Reset Classification" in Python API but I don't see a function to perform this action. So, I used "Created (never classified)" as source points. But output is not what I expected it to be.