I did three experiments on the benchmark model bev_depth_lss_r50_256×704_128×128_24e_2key.py,but my map is only 0.30 at the highest and NDS is 0.37 at the highest,the result you reported in the program is map:0.33; NDS:0.43 exists a big distance,I don't know how to get the expected result. I would like to know if the author has any training tips?
I am training on a 3090Ti with bachsize=8, thank you very much if the author can answer this!
I did three experiments on the benchmark model bev_depth_lss_r50_256×704_128×128_24e_2key.py,but my map is only 0.30 at the highest and NDS is 0.37 at the highest,the result you reported in the program is map:0.33; NDS:0.43 exists a big distance,I don't know how to get the expected result. I would like to know if the author has any training tips? I am training on a 3090Ti with bachsize=8, thank you very much if the author can answer this!