是否训练数据需要LIDAR数据辅助深度监督？

[zhaoyiming]

我在论文中没有找到需要激光雷达数据的描述，但是我在配置文件中发现了LoadPointsFromFile和MultiScaleDepthMapGenerator加载流程。能否帮忙确认一下这个工作需要额外激光雷达点监督？如果是，请标注一下，这可能有助于工作者快速筛选。
train_pipeline = [
dict(type="LoadMultiViewImageFromFiles", to_float32=True),
# dict(
# type="LoadPointsFromFile",
# coord_type="LIDAR",
# load_dim=5,
# use_dim=5,
# file_client_args=file_client_args,
# ),
dict(type="ResizeCropFlipImage"),
# dict(
# type="MultiScaleDepthMapGenerator",
# downsample=strides[:num_depth_layers],
# ),
dict(type="BBoxRotation"),
dict(type="PhotoMetricDistortionMultiViewImage"),
dict(type="NormalizeMultiviewImage", **img_norm_cfg),
dict(
type="CircleObjectRangeFilter",
class_dist_thred=[55] * len(class_names),
),
dict(type="InstanceNameFilter", classes=class_names),
dict(type="NuScenesSparse4DAdaptor"),
dict(
type="Collect",
keys=[
"img",
"timestamp",
"projection_mat",
"image_wh",
"gt_depth",
"focal",
"gt_bboxes_3d",
"gt_labels_3d",
],
meta_keys=["T_global", "T_global_inv", "timestamp", "instance_id"],
),
]

[zhaoyiming]

谢谢

[nschi00]

In the SparseDrive repository, the author commented that lidar depth supervision stabilizes the training process but you can also train the model without it, when lowering the learning rate.