The selection you've made does indeed show some of the best designs in terms of accuracy. Though some of them still have some minor or medium issues if you wanna consider them accurate.
The most accurate in the game, might just be Sinosauropteryx. Not just because of the accurate feathering. But also the accurate colouring. And honestly, Utahraptor might be up there. But it's hard to tell since we haven't really seen it in action yet. But it seems like the feathering on the hands might be correct. Which is a problem many of the other designs still get wrong. Gigantoraptor and Oviraptor still have the problem that their hands are entirely bald. So they don't have correct primary feathers. Don't get me wrong, besides that, they're still really accurate for the most part.
The Concavenator looks really quite accurate. We'll have to see how big it is in the game, since the actual animal is really not that big. There's no evidence of quills on the back of the neck. And it's hard to tell how the arms would have looked. There are bumps on the ulna. Those might be quillknobs. In which case, they could be quills like shown here. Or perhaps it might have had more feathering than that. It's hard to tell. What's a really nice touch, is the wide scute like scales on the bottom of the neck. This is something we know Concavenator has on the bottom of it's tail. And we know Allosaurus also has these on the bottom of its neck. So it seems reasonable to put them on the neck of Conca here as well.
Yutyrannus is genuinely a great design in terms of accuracy. It's hard to tell what the arms would have looked like. Since it doesn't have proper wings like dromaeosaurs or oviraptorosaurs have, I'm not sure if it's accurate to still show them bald like shown here. But it's definitely less of a problem because of the sparse feathering on the arms.
Deinocheirus is also really good. But, it might actually be too heavily feathered. It's a very large bulky animal that might have had trouble cooling down if it was so heavily feathered. so perhaps a lighter coat of feathers might be better. But overall, really good.
Jeholopterus is very good in terms of accuracy. But it's a bit more speculative. We know pterosaurs were fuzzy. But we don't know if they had fuzzy display structures like shown here.
Lystrosaurus isn't entirely accurate. It's a nice design. But they've made changes to the skull shape.
Archelon, Kronosaurus and Shonisaurus are all really good as far as I know. Although it's quite likely that Shonisaurus would have had a dorsal fin.
Dunkelosteus looks great. And I love how it doesn't look like an ugly monster, but an actual animal. The face looks really quite accurate. But it has one big problem. It's much too long. A recent paper showed that the classic style of Dunk reconstructions most likely are not correct. So it would have roughly been 2/3s of the length it is here.
Minmi and Wuerhosaurus I wouldn't consider that accurate. They're cool designs. But both of them have stylistic choices that don't really fit the animal if it's just accuracy we're after.
Australovenator also, I wouldn't put here on this list. The skull has some odd stylistic choices that aren't accurate and the hand claws are much too small. Megaraptid hand claws are massive. And a mistake that many paleoartist also make is that they just look at the bony claws. When in life, the keratin would have extended the length considerably. Around 30% more length in some cases.
Styxosaurus is cool. But it's VERY speculative. It's not impossible that it might have display structures like that. But I dunno how likely that would be.
The feathered T. rex is really cool. And it seems likely that it would have had some slight fuzziness to it. Funnily enough, the feathers shown here are probably at like the upper end of what would be plausible for T. rex. Other than that, it's still the JP T. rex underneath. Which is a fantastic design. Mostly fairly accurate. But with some obvious stylistic choices. And, it should probably have lips to cover the teeth.
Quetzalcoatlus is rather good. But it's much, too big and some of the proportions are off.
Coelophysis is pretty good. But the real animal would probably have been feathered a bit. Other fairly basal theropods like Dilophosaurus hve been found to most likely have a fair bit of feather covering. So it seems unlikely that Coelophysis would have been entirely bald.
Speaking of Dilophosaurus, Cryolophosaurus is a close relative of it. And it would also likely have had some feathering.
Suchomimus is definitely the most accurate spinosaurid in the game. It's quite good too. There's some minor proportional issues. And the skull still seems to follow the old reconstruction. The real skull would have been a little more robust in the back. For spinosaurids it's not entirely clear if they would have feathers. We know of fairly extensive scales in Allosauroids and Abelisaurids. So it's not unreasonable that spinosaurids might have been mostly, or entirely scaly. Like other theropods tho, it would likely have had lips to cover the teeth. And the hand claws are also too small.
Overall, this is a really good selection with a bunch of designs that are really accurate. There's some other designs that are quite good as well. Some that spring to mind are: Albertosaurus, Attenborosaurus, Geosternbergia, Ceratosaurus, JP3 Brachiosaurus, Frontier Giganotosaurus, Frontier Iguanodon, Euoplocephalus, Dsungaripterus, Corythosaurus, JWD Parasaurolophus
Some of them have minor proportional issues or lack feathers when they should. But overall they're quite accurate.