筛选检测结果

Detections 类提供的先进筛选功能，使您能够高效地缩小和优化目标检测结果。本节将介绍各种筛选方法，包括按特定类别或一组类别、置信度、目标面积、边界框面积、相对面积、框尺寸以及指定区域进行筛选。每种方法都配有简洁的代码示例，以帮助用户清晰地了解如何在应用中实现这些筛选器。

按特定类别筛选

允许您仅选择属于一个已选类别的检测结果。

筛选后筛选前

import supervision as sv

detections = sv.Detections(...)
detections = detections[detections.class_id == 0]

import supervision as sv

detections = sv.Detections(...)
detections = detections[detections.class_id == 0]

按类别集合筛选

允许您仅选择属于已选类别集合的检测结果。

筛选后筛选前

import numpy as np
import supervision as sv

selected_classes = [0, 2, 3]
detections = sv.Detections(...)
detections = detections[np.isin(detections.class_id, selected_classes)]

import numpy as np
import supervision as sv

class_id = [0, 2, 3]
detections = sv.Detections(...)
detections = detections[np.isin(detections.class_id, class_id)]

按置信度筛选

允许您选择具有特定置信度的检测结果，例如高于选定阈值的检测结果。

筛选后筛选前

import supervision as sv

detections = sv.Detections(...)
detections = detections[detections.confidence > 0.5]

import supervision as sv

detections = sv.Detections(...)
detections = detections[detections.confidence > 0.5]

按面积筛选

允许您根据检测结果的大小进行选择。我们将面积定义为检测结果在图像中占据的像素数量。在下面的示例中，我们筛选掉了过小的检测结果。

筛选后筛选前

import supervision as sv

detections = sv.Detections(...)
detections = detections[detections.area > 1000]

import supervision as sv

detections = sv.Detections(...)
detections = detections[detections.area > 1000]

按相对面积筛选

允许您根据检测结果相对于整个图像大小的比例来选择。有时，检测结果的大小概念会因图像而异。在 1280x720 图像上占据 10000 像素的检测结果可能很大，但在 3840x2160 图像上可能很小。在这种情况下，我们可以根据检测结果占据图像面积的百分比进行筛选。在下面的示例中，我们移除了过大的检测结果。

筛选后筛选前

import supervision as sv

image = ...
height, width, channels = image.shape
image_area = height * width

detections = sv.Detections(...)
detections = detections[(detections.area / image_area) < 0.8]

import supervision as sv

image = ...
height, width, channels = image.shape
image_area = height * width

detections = sv.Detections(...)
detections = detections[(detections.area / image_area) < 0.8]

按边界框尺寸筛选

允许您根据检测结果的尺寸进行选择。边界框的大小以及其坐标都可以作为拒绝检测标准的依据。实现此类筛选需要一些自定义代码，但相对简单快捷。

筛选后筛选前

import supervision as sv

detections = sv.Detections(...)
w = detections.xyxy[:, 2] - detections.xyxy[:, 0]
h = detections.xyxy[:, 3] - detections.xyxy[:, 1]
detections = detections[(w > 200) & (h > 200)]

import supervision as sv

detections = sv.Detections(...)
w = detections.xyxy[:, 2] - detections.xyxy[:, 0]
h = detections.xyxy[:, 3] - detections.xyxy[:, 1]
detections = detections[(w > 200) & (h > 200)]

按 PolygonZone 筛选

允许您将 Detections 与 PolygonZone 结合使用，以剔除区域内外的边界框。在下面的示例中，您可以看到如何滤除图像下半部分的所有检测结果。

筛选后筛选前

import supervision as sv

zone = sv.PolygonZone(...)
detections = sv.Detections(...)
mask = zone.trigger(detections=detections)
detections = detections[mask]

import supervision as sv

zone = sv.PolygonZone(...)
detections = sv.Detections(...)
mask = zone.trigger(detections=detections)
detections = detections[mask]

按混合条件筛选

然而，Detections 最强大的地方在于，你可以通过简单地使用 & 或 | 组合单独的条件，来构建任意复杂的逻辑条件。

筛选后筛选前

import supervision as sv

zone = sv.PolygonZone(...)
detections = sv.Detections(...)
mask = zone.trigger(detections=detections)
detections = detections[(detections.confidence > 0.7) & mask]

import supervision as sv

zone = sv.PolygonZone(...)
detections = sv.Detections(...)
mask = zone.trigger(detections=detections)
detections = detections[mask]

Comments