megengine.functional.nn.nms¶

nms(boxes, scores, iou_thresh, max_output=None)[source]¶

Performs non-maximum suppression (NMS) on the boxes according to their intersection-over-union(IoU).

Parameters

boxes (Tensor) – tensor of shape (N, 4); the boxes to perform nms on; each box is expected to be in (x1, y1, x2, y2) format.
iou_thresh (float) – IoU threshold for overlapping.
scores (Tensor) – tensor of shape (N,), the score of boxes.
max_output (Optional[int]) – the maximum number of boxes to keep; it is optional if this operator is not traced otherwise it required to be specified; if it is not specified, all boxes are kept.

Return type

Tensor

Returns

indices of the elements that have been kept by NMS, sorted by scores.

Note

max_output should be specified and should have valid positive value under tracing.

Examples

>>> import numpy as np
>>> x = np.zeros((100,4))
>>> np.random.seed(42)
>>> x[:,:2] = np.random.rand(100,2)*20
>>> x[:,2:] = np.random.rand(100,2)*20 + 100
>>> scores = Tensor(np.random.rand(100))
>>> inp = Tensor(x)
>>> F.vision.nms(inp, scores, iou_thresh=0.7)
Tensor([75 69], dtype=int32, device=xpux:0)

megengine.functional.nn.roi_align

megengine.functional.nn.correlation