Clouds and nighttime are a barrier to visual detection. Even with good effectiveness the conditions needed for that would mean that you have far less than 50% uptime, and your downtime is predictable to your adversary.
A cheap radar takes an order of magnitude less power to run on hardware that is cheaper than an LLM and can see way farther than a camera.
Radar can't effectively see the new low-altitude small fiber optic spool first-person drones that are redefining frontline war.
The solution is going to be multi-modal (optical + audio) and imperfect. The above poster is correct that an ordinary camera with computer vision (not an LLM, of course) is going to be part of the multimodal defense in depth.
Shahed-type one-way attack drones are important to defend against, but not as impactful in terms of frontline body count, given they just slam into a pre-programmed target.