ProposalRecall

class mmeval.metrics.ProposalRecall(iou_thrs: Optional[Union[float, Sequence[float]]] = None, proposal_nums: Union[int, Sequence[int]] = (1, 10, 100, 1000), use_legacy_coordinate: bool = False, nproc: int = 4, print_results: bool = True, **kwargs)

Proposals recall evaluation metric.

The speed of calculating recall is faster than COCO Detection metric.

参数

• iou_thrs (float | List[float], optional) – IoU thresholds. If not specified, IoUs from 0.5 to 0.95 will be used. Defaults to None.

• proposal_nums (Sequence[int]) – Numbers of proposals to be evaluated. Defaults to (1, 10, 100, 1000).

• use_legacy_coordinate (bool) – Whether to use coordinate system in mmdet v1.x. which means width, height should be calculated as ‘x2 - x1 + 1` and ‘y2 - y1 + 1’ respectively. Defaults to False.

• nproc (int) – Processes used for computing TP and FP. If nproc is less than or equal to 1, multiprocessing will not be used. Defaults to 4.

• print_results (bool) – Whether to print the results. Defaults to True.

• **kwargs – Keyword parameters passed to BaseMetric.

实际案例

>>> import numpy as np
>>> from mmeval import ProposalRecall
>>>
>>> proposal_recall = ProposalRecall()
>>> def _gen_bboxes(num_bboxes, img_w=256, img_h=256):
...     # random generate bounding boxes in 'xyxy' formart.
...     x = np.random.rand(num_bboxes, ) * img_w
...     y = np.random.rand(num_bboxes, ) * img_h
...     w = np.random.rand(num_bboxes, ) * (img_w - x)
...     h = np.random.rand(num_bboxes, ) * (img_h - y)
...     return np.stack([x, y, x + w, y + h], axis=1)
>>>
>>> prediction = {
...     'bboxes': _gen_bboxes(10),
...     'scores': np.random.rand(10, ),
... }
>>> groundtruth = {
...     'bboxes': _gen_bboxes(10),
... }
>>> proposal_recall(predictions=[prediction, ], groundtruths=[groundtruth, ])  
{'AR@1': ..., 'AR@10': ..., 'AR@100': ..., 'AR@1000': ...}

add(predictions: Sequence[Dict], groundtruths: Sequence[Dict]) → None

Add the intermediate results to self._results.

参数

• predictions (Sequence[dict]) –

  A sequence of dict. Each dict representing a detection result for an image, with the following keys:

  • bboxes (numpy.ndarray): Shape (N, 4), the predicted bounding bboxes of this image, in ‘xyxy’ foramrt.

  • scores (numpy.ndarray): Shape (N, 1), the predicted scores of bounding boxes.

• groundtruths (Sequence[dict]) –

  A sequence of dict. Each dict represents a groundtruths for an image, with the following keys:

  • bboxes (numpy.ndarray): Shape (M, 4), the ground truth bounding bboxes of this image, in ‘xyxy’ foramrt.

calculate_recall(predictions: List[dict], groundtruths: List[dict], pool: Optional[multiprocessing.pool.Pool])

Calculate recall.

参数

• predictions (List[dict]) – A list of dict. Each dict is the detection result of an image. Same as ProposalRecall.add.

• groundtruths (List[dict]) – A list of dict. Each dict is the ground truth of an image. Same as ProposalRecall.add.

• pool (Optional[Pool]) – A instance of multiprocessing.Pool. If None, do not use multiprocessing.

返回

Shape (len(proposal_nums), len(iou_thrs)), the recall results.

返回类型

numpy.ndarry

compute_metric(results: list) → dict

Compute the ProposalRecall metric.

参数

results (List[tuple]) – A list of tuple. Each tuple is the prediction and ground truth of an image. This list has already been synced across all ranks.

返回

The computed metric. The keys are the names of the proposal numbers, and the values are corresponding results.

返回类型

dict

process_proposals(predictions: List[dict], groundtruths: List[dict]) → Tuple[List, List, int]

Process the proposals(bboxes) in predictions and groundtruths.

参数

• predictions (List[dict]) – A list of dict. Each dict is the detection result of an image. Same as ProposalRecall.add.

• groundtruths (list[dict]) – Same as ProposalRecall.add.

返回

• proposal_preds (List[numpy.ndarray]): The sorted proposals of the prediction.

• proposal_gts (List[numpy.ndarray]): The proposals of the groundtruths.

• num_gts (int): The total groundtruth numbers.

返回类型

tuple (proposal_preds, proposal_gts, num_gts)