Don't overestimate the power of radar. If we could detect a stealth bomber from across the planet, that's still only about 12k kilometers. Similar shielding is mostly a matter of geometry. If you can make a weapon that can launch that much mass at that speed, you can manage some carbon fiber composite to cover it, and that's not even counting on more advanced
metamaterials which could make it literally invisible.
Mars is more like 71,950k kilometers away, at absolute minimum, or four light minutes. A relativistic weapon would have to be launched from much further. The asteroid belt would be a good platform for an explosive-launched projectile, and that's eight light minutes away (again, at minimum), 20 AU (a butt-load of kilometers) in circumference. It would provide lots of interferences to mess with the data further. Pluto is 5.5 light hours, I think it was. That's less than five and a half hours to scramble something to intersect something that could be coming from any direction, since once you're working with relativistic weapons at that distance, pushing a kuiper belt object closer would be trivial, and that's if you're not launching from further. At that circumference you're looking at 245 AU of coverage. If you want that five and a half hour coverage from any direction (a defense against other systems), you'd need to cover 427 square AUs with sensors if my math is right, and from larger distances, faster projectiles get a lot easier to launch, so you'd have less warning on a faster moving object. Something like this would probably eclipse a Dyson sphere in difficulty in maintenance and power support alone, since you'd probably have to set off a couple of nukes for each radar pulse to have any chance of detecting a shielded weapon.
For that matter, telescopes aren't much better equipped. You have exposure times limiting the distance at which they're useful, which isn't something you can really hand-wave away except with very, very large telescopes. Maybe possible with lighter-weight nanomaterials as collectors, I admit, but it's still going to be huge. You need to aim the telescopes at each useful arc-n of space (second, millisecond, whatever depending on aperture size) You probably want to make three big spherical ones, because every celestial body is going to provide an ever-expanding conical blindspot otherwise, particularly the sun. You need at least two looking at the same object at pretty good distances apart to reliably determine speed. Then you need to process an unimaginable amount of data compared to what we can do now, filter out artifacts, decide on the distance of any object apparently fast enough to pose a threat (we can't ignore so much as a speck of dust floating past one of the telescopes nearby), and finally filter out all of the false positives by doing it again a couple of times just to be sure it wasn't a prank. Add to it that visible stealth (which isn't terribly difficult, we have a pigment now that
absorbs 99.9 percent visible light), and it gets even harder. It's not as hopeless for this, maybe, because we don't really know what sort of computers we'll have available at this point, but you still have the same problem of scrambling something fast enough once you do see a projectile coming.
TLDR: Space is beeg, and if anyone gets relativistic weapons, we're probably toast.