Add Image masking

[travisdriver]

Added an optional mask to the Image class, and also added functionality for filtering extracted keypoints with respect to this mask.

An example where this may be useful are when images are consistently obstructed (e.g., a spacecraft rendezvous with the ISS, landing on the Moon).

Moreover, from a previous project, it has also been shown that masking out dark regions (pixel value of ~0) of images captured during small body relative navigation scenarios (e.g., the interface between the foreground body and the background deep space, self-shadowing) substantially increases the reliability of the visual measurements:

No masking

Masking

[akshay-krishnan]

I think John is addressing most concerns I had, I am also not very familiar with the frontend APIs. I can take a final look once all other comments are addressed, if necessary.

[johnwlambert]

@travisdriver thanks for the updates. Can you add the masked and unmasked jobs to the CI like we discussed in our last meeting? That way we can track improvements even without the mask, and we can ensure we retain good performance on the masked job without degradation

[travisdriver]

@travisdriver thanks for the updates. Can you add the masked and unmasked jobs to the CI like we discussed in our last meeting? That way we can track improvements even without the mask, and we can ensure we retain good performance on the masked job without degradation

I'd like to save that for a different PR if possible, as I want to change a couple things about the AstroNet benchmarks.

[johnwlambert]

@travisdriver thanks for the updates. Can you add the masked and unmasked jobs to the CI like we discussed in our last meeting? That way we can track improvements even without the mask, and we can ensure we retain good performance on the masked job without degradation

I'd like to save that for a different PR if possible, as I want to change a couple things about the AstroNet benchmarks.

@travisdriver Sure thing, can you open an issue so we don't lose track of it later?

Could you also upload the visual dashboard file here in the comments, so we can see the diff?

[johnwlambert]

@travisdriver thanks for the updates. Can you add the masked and unmasked jobs to the CI like we discussed in our last meeting? That way we can track improvements even without the mask, and we can ensure we retain good performance on the masked job without degradation

I'd like to save that for a different PR if possible, as I want to change a couple things about the AstroNet benchmarks.

@travisdriver Sure thing, can you open an issue so we don't lose track of it later?

Could you also upload the visual dashboard file here in the comments, so we can see the diff?

Hi @travisdriver, I think this PR is almost ready to merge : - )
Just these last two things here above

[ayushbaid]

nit: len(self) looks better. I am fine with this too.

[johnwlambert]

dashboard:
visual_comparison_dashboard.html 2.zip