Forward function. including multi-level image features, text features: BxLxD

Forward function.

Forward features from the upstream network.

Forward features from the upstream network.

Forward function of contrastive learning.

Forward function of contrastive learning.

Forward function of contrastive learning.

Forward features from the upstream network.

Forward features from the upstream network.

Perform normalization, padding and bgr2rgb conversion based on ``DetDataPreprocessorr``. Args: data (dict): Data sampled

Forward process.

Forward process.

Forward process.

Forward process.

Forward process.

Forward process.

Forward process.

Forward process.

Forward function of loss.

Assign gt to bboxes. The assignment is done in following steps 1. compute alignment metric between all bbox (bbox of all pyramid

Forward feature of a single scale level.

Forward feature of a single scale level.

Forward features from the upstream network.

Get annotation by index.

Call function to collect indexes. Args: dataset (:obj:`Dataset` or list): The dataset or cached list. Returns:

Call function to collect indexes. Args: dataset (:obj:`Dataset` or list): The dataset or cached list. Returns:

Call function to collect indexes. Args: dataset (:obj:`Dataset` or list): The dataset or cached list. Returns:

Call function to collect indexes. Args: dataset (:obj:`Dataset` or list): The dataset or cached list. Returns:

Initialize the weight and bias of PPYOLOE head.

Initialize the weight and bias of PPYOLOE head.

Load annotations from an annotation file named as ``self.ann_file`` Returns: List[dict]: A list of annotation.

Load annotations from an annotation file named as ``self.ann_file`` Returns: List[dict]: A list of annotation.

Load annotations from an annotation file named as ``self.ann_file`` Returns: List[dict]: A list of annotation.

Perform forward propagation and loss calculation of the detection head on the features of the upstream network.

Perform forward propagation and loss calculation of the detection head on the features of the upstream network.

Calculate losses from a batch of inputs and data samples.

Calculate losses from a batch of inputs and data samples.

Perform forward propagation of the head, then calculate loss and predictions from the features and data samples.

Perform forward propagation of the head, then calculate loss and predictions from the features and data samples.

Mixed image data transformation. Args: results (dict): Result dict. Returns: results (dict): Updated result

Mixed image data transformation. Args: results (dict): Result dict. Returns: results (dict): Updated result

YOLOv5 MixUp transform function. Args: results (dict): Result dict Returns: results (dict): Updated result d

YOLOX MixUp transform function. Args: results (dict): Result dict. Returns: results (dict): Updated result d

Perform forward propagation of the detection head and predict detection results on the features of the upstream network.

Perform forward propagation of the detection head and predict detection results on the features of the upstream network.

Predict results from a batch of inputs and data samples with post- processing.

Predict results from a batch of inputs and data samples with post- processing.

Since YOLO series algorithms will inherit from YOLOv5Head, but different algorithms have special initialization process. The special_

Convert the model into training mode while keep normalization layer frozen.

Convert the model into training mode while keep normalization layer frozen.

Data augmentation function. The transform steps are as follows: 1. Randomly generate index list of other images. 2. Before Mo

Rewrite collate_fn to get faster training speed. Args: data_batch (Sequence): Batch of data. use_ms_training (bool): Whether to use