Masks - How and where are they used in processing?

Hi All - I have a dataset that requires some masking to eliminate unwanted noise/etc in the point cloud and subsequent ortho. Once the filename_mask files are added to the images, where are they used in the processing? Specifically, I am wondering if there is a way that the “masked” areas can be used for features and matching…but not in subsequent processing steps (e.g. reconstruction/pc generation)? Thank you in advance for any feedback on this.

1 Like

Really interesting question…

My understanding of the pipeline might be wrong, but I think that the Masks apply to ortho/point cloud only, not matching.

Don’t quote me on that yet, though, haha.

Thank you for the feedback! The masks seem present in the heatmap…which is making me think they are applied earlier in the process unfortunately…I will keep digging, but any thoughts relative to applying masking later in the process would be greatly appreciated!

Thanks again!

1 Like

Yeah, you certainly may be correct. My understanding is pretty limited in this arena, so thank you for giving me feedback.

Just to clarify, there are features you might want to match off of, that you wouldn’t necessarily want in the final reconstruction, right? Is that the intent for having the mask apply AFTER matching?

That is correct. I was really hoping the unmasked information could be used up to reconstruction, but with the masks applied and masked data/areas not utilized after…

1 Like

Feature Request: Option to apply Image Masks After Matching · Issue #1303 · OpenDroneMap/ODM (

I’ve opened a Feature Request to track this.

Please feel free to provide your thoughts or examples where this change would help here or there.


This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.