Adaptive Spatio-temporal Sampling

Adaptive Mixing

An attention module used in BEVFormer based on Deformable-Detr. `Deformable DETR: Deformable Transformers for End-to-End Object Detection.

Implement the scaled dot product attention with softmax. Arguments --------- softmax_scale: The temperature to use for the softmax att

r"""Criterion that computes Focal loss. According to :cite:`lin2018focal`, the Focal loss is computed as follows: .. math:: \text{FL}(

Apply photometric distortion to image sequentially, every transformation is applied with a probability of 0.5. The position of random contrast is

Model Exponential Moving Average from https://github.com/rwightman/ pytorch-image-models Keep a moving average of everything in the model stat

Adaptive Spatio-temporal Sampling

Adaptive Spatio-temporal Sampling

Scale-adaptive Self Attention

BBox3DL1Cost. Args: weight (int | float, optional): loss_weight

r"""Criterion that computes Focal loss. According to :cite:`lin2018focal`, the Focal loss is computed as follows: .. math:: \text{FL}(

Save checkpoints periodically. Args: interval (int): The saving period. If ``by_epoch=True``, interval indicates epochs, othe

r"""Feature Pyramid Network. LSSFPN: 1/16 + 1/32 -> lateral -> 1/32 upsample to 1/16 + 1/16 -> conv This is an implementation of paper `F

Args: num_layer (List[int]): ResBlock for each layer num_channels (List[int]): Output channel number for each stage

EMAHook used in BEVDepth. Modified from https://github.com/Megvii-Base Detection/BEVDepth/blob/main/callbacks/ema.py.

r"""Lift-Splat-Shoot view transformer with BEVPoolv2 implementation. Please refer to the `paper <https://arxiv.org/abs/2008.05711>`_ and

Load Points From File. Load points from file. Args: coord_type (str): The type of coordinates of points cloud. Available

Load Points From File. Load points from file. Args: coord_type (str): The type of coordinates of points cloud. Available

Load radar Points From File. Load sunrgbd and scannet points from file. Args: coord_type (str): The type of coordinates of points cl

EMAHook used in BEVDepth. Modified from https://github.com/Megvii-Base Detection/BEVDepth/blob/main/callbacks/ema.py.

Bbox coder for NMS-free detector. Args: pc_range (list[float]): Range of point cloud. post_center_range (list[float]): Limit of th

Normalize the image. Added key is "img_norm_cfg". Args: mean (sequence): Mean values of 3 channels. std (sequence): Std values

Pad the multi-view image. There are two padding modes: (1) pad to a fixed size and (2) pad to the minimum size that is divisible by some numbe

Apply photometric distortion to image sequentially, every transformation is applied with a probability of 0.5. The position of random contrast is

r"""BEVPoolv2 implementation for Lift-Splat-Shoot view transformation. Please refer to the `paper <https://arxiv.org/abs/2211.17111>`_

Default formatting bundle. It simplifies the pipeline of formatting common fields for voxels, including "proposals", "gt_bboxes", "gt_labels"

Flip the image & bbox & mask. If the input dict contains the key "flip", then the flag will be used, otherwise it will be randomly decided by

BBox3DL1Cost. Args: weight (int | float, optional): loss_weight

Normalize the image. Added key is "img_norm_cfg". Args: mean (sequence): Mean values of 3 channels. std (sequence): Std valu

Monocular 3D detection on VoD Dataset. Args: data_root (str): Path of dataset root. info_file (str): Path of info file. l