Detection ability drops significantly at long distances

[zizhengu]

When I used DSVT on my own dataset, I found that it has excellent detection ability for nearby objects (0-30m), but the mean Average Precision drops significantly at long distances (30-60m,60m-inf) compared to Centerpoints model. Can we explain it?

[Haiyang-W]

I'm not sure about that. It makes sense that our backbone and sparse conv perform equally well at long and short distances, at least from what we've seen on waymo. I suspect it's because of your previous parameter error issue.