【深度学习】【目标检测】指标计算

最新推荐文章于 2024-10-10 15:21:36 发布

Hanawh

最新推荐文章于 2024-10-10 15:21:36 发布

阅读量627

点赞数

分类专栏：深度学习文章标签：深度学习

本文链接：https://blog.csdn.net/qq_36530992/article/details/107024155

版权

本文探讨了在深度学习目标检测中，如何计算预测框与 ground truth 的 IoU，选择最大 IoU 对应的 gt，并针对多个 IoU 阈值计算 PR 曲线。通过置信度排序，计算类别级别的 Precision 和 Recall，进而求得平均 Precision (AP)。最终，文章提及在置信度为0.1时的 Precision 和 Recall 预估值。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

def ap_per_class(tp, conf, pred_cls, target_cls):
    """ Compute the average precision, given the recall and precision curves.
    Source: https://github.com/rafaelpadilla/Object-Detection-Metrics.
    # Arguments
        tp:    True positives (nparray, nx1 or nx10). 
        conf:  Objectness value from 0-1 (nparray).
        pred_cls: Predicted object classes (nparray).
        target_cls: True object classes (nparray).
    # Returns
        The average precision as computed in py-faster-rcnn.
    """
    '''
    tp : [bt*300, 10]
    conf : [bt*300, ]
    pred_cls : [bt*300, ]
    target_cls : [num_gt,]
    '''

    # Sort by objectness
    i = np.argsort(-conf)
    tp, conf, pred_cls = tp[i], conf[i], pred_cls[i]

    # Find unique classes
    unique_classes = np.unique(target_cls)

    # Create Precision-Recall curve and compute AP for each class
    pr_score = 0.1  # score to evaluate P and R https://github.com/ultralytics/yolov3/issues/898
    s = [unique_classes.shape[0], tp.shape[1]]  # number class, number iou thresholds (i.e. 10 for mAP0.5...0.95)
    ap, p, r = np.zeros(s), np.zeros(s), np.zeros(s)
    <