ilnehc

@Eralien Thank you for raising the issue. Could you provide the segment or json number that we can look into? Btw, I notice that you have an issue related to...

@Eralien We are working on it. Due to the large size of the dataset and limited time with other projects on hand, a huge effort is needed to make sure...

@Mollylulu As said in the paper, 'with the help of the localization and mapping system, 3D lane points in frames within a segment could be spliced into long, high-density lanes'.

@messier202 Thank you for raising this information that we forgot to clarify. It should be 30 & 0.3. However, we found some misaligned results recently. We will update them and...

@callzhang Please see the related issue above. This checkpoint is outdated and we are not allowed to provide a pretrained model.

Please see https://github.com/OpenPerceptionX/OpenLane/issues/15, and we will update the eval script accordingly.

@zimurui Hi, could you elaborate more cause I did not fully understand your question. LaneATT predicts the length of lanes directly while our 2D branch predicts each row with visibility...

@zimurui Yes, it is different from the original LaneATT and we did not try it. We design this to make 2D/3D branches consistent at the very beginning. Intuitively, the critical...

@usherbob Just change OpenPerceptionX to OpenDriveLab: https://github.com/OpenDriveLab/OpenLane/blob/main/anno_criterion/Lane/README.md

@adilsonmedronha Thanks. Please refer to https://github.com/OpenDriveLab/PersFormer_3DLane/issues/8.