The ethical argument for why everyone should have access to cryptography is a lot stronger than why everyone should have access to satellite imagery recognition algorithms.
Also, cryptography requires both sides to use the same algorithm, while companies don't need to use the same recognition algorithms.
It also helped, in the crypto case, that you could print some version of it on a T-shirt or mail it on a postcard. It looked like speech, while neural net parameters don't.
Lawyers have made strong cases on both sides of the cryptography argument; probably they can on both sides of the satellite-imagery argument as well. Maps are the primary result of satellite imagery recognition and are a public good. Most covert activity visible on satellite imagery is environmental damage, which is often illegal and generally harms the public. Satellite image processing can be very useful for increasing agricultural production; restricting that to one country, or granting one country's companies an effective worldwide monopoly on increasing agricultural production, would be ethically unconscionable — in times of drought, it amounts to letting people starve instead of telling them how to raise adequate food.
But Bernstein's case didn't hinge on the likely consequences of strong cryptography being widely available; rather, he argued that he had a First Amendment right to publish his research.
I believe the emphasis here was on the generic right of disseminating research not on judging the necessity of a specific technology for particular audience.
Also, cryptography requires both sides to use the same algorithm, while companies don't need to use the same recognition algorithms.
It also helped, in the crypto case, that you could print some version of it on a T-shirt or mail it on a postcard. It looked like speech, while neural net parameters don't.
So the free speech case seems much weaker.