Python程序实现NMS

NMS的算法步骤如下：

# INPUT：所有预测出的bounding box (bbx)信息（坐标和置信度confidence），　IOU阈值（大于该阈值的bbx将被移除）
for object in all objects:
	(1) 获取当前目标类别下所有bbx的信息
	(2) 将bbx按照confidence从高到低排序,并记录当前confidence最大的bbx
	(3) 计算最大confidence对应的bbx与剩下所有的bbx的IOU,移除所有大于IOU阈值的bbx
	(4) 对剩下的bbx，循环执行(2)和(3)直到所有的bbx均满足要求（即不能再移除bbx）

需要注意的是，NMS是对所有的类别分别执行的。举个栗子，假设最后预测出的矩形框有2类（分别为cup, pen）,在NMS之前，每个类别可能都会有不只一个bbx被预测出来，这个时候我们需要对这两个类别分别执行一次NMS过程。
我们用python编写NMS代码，假设对于一张图片，所有的bbx信息已经保存在一个字典中，保存形式如下：
 

predicts_dict: {"cup": [[x1_1, y1_1, x2_1, y2_1, scores1], [x1_2, y1_2, x2_2, y2_2, scores2], ...], "pen": [[x1_1, y1_1, x2_1, y2_1, scores1], [x1_2, y1_2, x2_2, y2_2, scores2], ...]}

即目标的位置和置信度用列表储存，每个列表中的一个子列表代表一个bbx信息。详细的代码如下： 

import numpy as np
def non_max_suppress(predicts_dict, threshold=0.2):
    """
    implement non-maximum supression on predict bounding boxes.
    Args:
        predicts_dict: {"stick": [[x1, y1, x2, y2, scores1], [...]]}.
        threshhold: iou threshold
    Return:
        predicts_dict processed by non-maximum suppression
    """
    for object_name, bbox in predicts_dict.items():   #对每一个类别的目标分别进行NMS
        bbox_array = np.array(bbox, dtype=np.float)
 
        ## 获取当前目标类别下所有矩形框（bounding box,下面简称bbx）的坐标和confidence,并计算所有bbx的面积
        x1, y1, x2, y2, scores = bbox_array[:,0], bbox_array[:,1], bbox_array[:,2], bbox_array[:,3], bbox_array[:,4]
        areas = (x2-x1+1) * (y2-y1+1)
        #print("areas shape = ", areas.shape)
 
        ## 对当前类别下所有的bbx的confidence进行从高到低排序（order保存索引信息）
        order = scores.argsort()[::-1]
        print("order = ", order)
        keep = [] #用来存放最终保留的bbx的索引信息
 
        ## 依次从按confidence从高到低遍历bbx，移除所有与该矩形框的IOU值大于threshold的矩形框
        while order.size > 0:
            i = order[0]
            keep.append(i) #保留当前最大confidence对应的bbx索引
 
            ## 获取所有与当前bbx的交集对应的左上角和右下角坐标，并计算IOU（注意这里是同时计算一个bbx与其他所有bbx的IOU）
            xx1 = np.maximum(x1[i], x1[order[1:]])　#当order.size=1时，下面的计算结果都为np.array([]),不影响最终结果
            yy1 = np.maximum(y1[i], y1[order[1:]])
            xx2 = np.minimum(x2[i], x2[order[1:]])
            yy2 = np.minimum(y2[i], y2[order[1:]])
            inter = np.maximum(0.0, xx2-xx1+1) * np.maximum(0.0, yy2-yy1+1)
            iou = inter/(areas[i]+areas[order[1:]]-inter)
            print("iou =", iou)
 
            print(np.where(iou<=threshold)) #输出没有被移除的bbx索引（相对于iou向量的索引）
            indexs = np.where(iou<=threshold)[0] + 1 #获取保留下来的索引(因为没有计算与自身的IOU，所以索引相差１，需要加上)
            print("indexs = ", type(indexs))
            order = order[indexs] #更新保留下来的索引
            print("order = ", order)
        bbox = bbox_array[keep]
        predicts_dict[object_name] = bbox.tolist()
        predicts_dict = predicts_dict
    return predicts_dict

在YOLO源代码general.py可见：

def non_max_suppression(
        prediction,
        conf_thres=0.25,
        iou_thres=0.45,
        classes=None,
        agnostic=False,
        multi_label=False,
        labels=(),
        max_det=300,
        nm=0,  # number of masks
):
    """Non-Maximum Suppression (NMS) on inference results to reject overlapping detections

    Returns:
         list of detections, on (n,6) tensor per image [xyxy, conf, cls]
    """

    # Checks
    assert 0 <= conf_thres <= 1, f'Invalid Confidence threshold {conf_thres}, valid values are between 0.0 and 1.0'
    assert 0 <= iou_thres <= 1, f'Invalid IoU {iou_thres}, valid values are between 0.0 and 1.0'
    if isinstance(prediction, (list, tuple)):  # YOLOv5 model in validation model, output = (inference_out, loss_out)
        prediction = prediction[0]  # select only inference output

    device = prediction.device
    mps = 'mps' in device.type  # Apple MPS
    if mps:  # MPS not fully supported yet, convert tensors to CPU before NMS
        prediction = prediction.cpu()
    bs = prediction.shape[0]  # batch size
    nc = prediction.shape[2] - nm - 5  # number of classes
    xc = prediction[..., 4] > conf_thres  # candidates

    # Settings
    # min_wh = 2  # (pixels) minimum box width and height
    max_wh = 7680  # (pixels) maximum box width and height
    max_nms = 30000  # maximum number of boxes into torchvision.ops.nms()
    time_limit = 0.5 + 0.05 * bs  # seconds to quit after
    redundant = True  # require redundant detections
    multi_label &= nc > 1  # multiple labels per box (adds 0.5ms/img)
    merge = False  # use merge-NMS

    t = time.time()
    mi = 5 + nc  # mask start index
    output = [torch.zeros((0, 6 + nm), device=prediction.device)] * bs
    for xi, x in enumerate(prediction):  # image index, image inference
        # Apply constraints
        # x[((x[..., 2:4] < min_wh) | (x[..., 2:4] > max_wh)).any(1), 4] = 0  # width-height
        x = x[xc[xi]]  # confidence

        # Cat apriori labels if autolabelling
        if labels and len(labels[xi]):
            lb = labels[xi]
            v = torch.zeros((len(lb), nc + nm + 5), device=x.device)
            v[:, :4] = lb[:, 1:5]  # box
            v[:, 4] = 1.0  # conf
            v[range(len(lb)), lb[:, 0].long() + 5] = 1.0  # cls
            x = torch.cat((x, v), 0)

        # If none remain process next image
        if not x.shape[0]:
            continue

        # Compute conf
        x[:, 5:] *= x[:, 4:5]  # conf = obj_conf * cls_conf

        # Box/Mask
        box = xywh2xyxy(x[:, :4])  # center_x, center_y, width, height) to (x1, y1, x2, y2)
        mask = x[:, mi:]  # zero columns if no masks

        # Detections matrix nx6 (xyxy, conf, cls)
        if multi_label:
            i, j = (x[:, 5:mi] > conf_thres).nonzero(as_tuple=False).T
            x = torch.cat((box[i], x[i, 5 + j, None], j[:, None].float(), mask[i]), 1)
        else:  # best class only
            conf, j = x[:, 5:mi].max(1, keepdim=True)
            x = torch.cat((box, conf, j.float(), mask), 1)[conf.view(-1) > conf_thres]

        # Filter by class
        if classes is not None:
            x = x[(x[:, 5:6] == torch.tensor(classes, device=x.device)).any(1)]

        # Apply finite constraint
        # if not torch.isfinite(x).all():
        #     x = x[torch.isfinite(x).all(1)]

        # Check shape
        n = x.shape[0]  # number of boxes
        if not n:  # no boxes
            continue
        x = x[x[:, 4].argsort(descending=True)[:max_nms]]  # sort by confidence and remove excess boxes

        # Batched NMS
        c = x[:, 5:6] * (0 if agnostic else max_wh)  # classes
        boxes, scores = x[:, :4] + c, x[:, 4]  # boxes (offset by class), scores
        i = torchvision.ops.nms(boxes, scores, iou_thres)  # NMS
        i = i[:max_det]  # limit detections
        if merge and (1 < n < 3E3):  # Merge NMS (boxes merged using weighted mean)
            # update boxes as boxes(i,4) = weights(i,n) * boxes(n,4)
            iou = box_iou(boxes[i], boxes) > iou_thres  # iou matrix
            weights = iou * scores[None]  # box weights
            x[i, :4] = torch.mm(weights, x[:, :4]).float() / weights.sum(1, keepdim=True)  # merged boxes
            if redundant:
                i = i[iou.sum(1) > 1]  # require redundancy

        output[xi] = x[i]
        if mps:
            output[xi] = output[xi].to(device)
        if (time.time() - t) > time_limit:
            LOGGER.warning(f'WARNING:NMS time limit {time_limit:.3f}s exceeded')
            break  # time limit exceeded

    return output

参考资料：目标检测中NMS（非极大抑制）的概念理解-CSDN博客