PackMultiTaskInputs

class mmpretrain.datasets.transforms.PackMultiTaskInputs(multi_task_fields, input_key='img', task_handlers={})

Convert all image labels of multi-task dataset to a dict of tensor.

参数:

• multi_task_fields (Sequence[str]) –

• input_key (str) –

• task_handlers (dict) –

transform(results)

Method to pack the input data.

result = {‘img_path’: ‘a.png’, ‘gt_label’: {‘task1’: 1, ‘task3’: 3},

‘img’: array([[[ 0, 0, 0])