YOLOv5/YOLOv8/YOLOv9改进——将损失函数改进为inner_iou、inner_giou、inner_diou、inner_ciou、inner_eiou、inner_siou

最新推荐文章于 2024-11-12 02:43:00 发布

长沙有肥鱼

最新推荐文章于 2024-11-12 02:43:00 发布

阅读量1.2k

点赞数 8

分类专栏：计算机视觉文章标签：深度学习计算机视觉 pytorch YOLO

本文链接：https://blog.csdn.net/weixin_53660567/article/details/139595162

版权

计算机视觉专栏收录该内容

19 篇文章

订阅专栏

💡 🎈🎄🎋🎍🎉🎑🎫🎞🎪🎁🎎🧵🛒🧶🥽🏅🥉🥇🥇🏅🏆🏆🥉

损失函数inner_iou

本文使用的YOLOv5、YOLOv8、YOLOv9版本均为最新版本。论文链接：https://arxiv.org/pdf/2311.02877.pdf

🎃代码修改

1.在mertric.py中将def bbox_iou

def bbox_iou(box1, box2, xywh=True, GIoU=False, DIoU=False, CIoU=False, eps=1e-7):
    # Returns Intersection over Union (IoU) of box1(1,4) to box2(n,4)

    # Get the coordinates of bounding boxes
    if xywh:  # transform from xywh to xyxy
        (x1, y1, w1, h1), (x2, y2, w2, h2) = box1.chunk(4, -1), box2.chunk(4, -1)
        w1_, h1_, w2_, h2_ = w1 / 2, h1 / 2, w2 / 2, h2 / 2
        b1_x1, b1_x2, b1_y1, b1_y2 = x1 - w1_, x1 + w1_, y1 - h1_, y1 + h1_
        b2_x1, b2_x2, b2_y1, b2_y2 = x2 - w2_, x2 + w2_, y2 - h2_, y2 + h2_
    else:  # x1, y1, x2, y2 = box1
        b1_x1, b1_y1, b1_x2, b1_y2 = box1.chunk(4, -1)
        b2_x1, b2_y1, b2_x2, b2_y2 = box2.chunk(4, -1)
        w1, h1 = b1_x2 - b1_x1, b1_y2 - b1_y1 + eps
        w2, h2 = b2_x2 - b2_x1, b2_y2 - b2_y1 + eps

    # Intersection area
    inter = (torch.min(b1_x2, b2_x2) - torch.max(b1_x1, b2_x1)).clamp(0) * \
            (torch.min(b1_y2, b2_y2) - torch.max(b1_y1, b2_y1)).clamp(0)

    # Union Area
    union = w1 * h1 + w2 * h2 - inter + eps

    # IoU
    iou = inter / union
    if CIoU or DIoU or GIoU:
        cw = torch.max(b1_x2, b2_x2) - torch.min(b1_x1, b2_x1)  # convex (smallest enclosing box) width
        ch = torch.max(b1_y2, b2_y2) - torch.min(b1_y1, b2_y1)  # convex height
        if CIoU or DIoU:  # Distance or Complete IoU https://arxiv.org/abs/1911.08287v1
            c2 = cw ** 2 + ch ** 2 + eps  # convex diagonal squared
            rho2 = ((b2_x1 + b2_x2 - b1_x1 - b1_x2) ** 2 + (b2_y1 + b2_y2 - b1_y1 - b1_y2) ** 2) / 4  # center dist ** 2
            if CIoU:  # https://github.com/Zzh-tju/DIoU-SSD-pytorch/blob/master/utils/box/box_utils.py#L47
                v = (4 / math.pi ** 2) * torch.pow(torch.atan(w2 / h2) - torch.atan(w1 / h1), 2)
                with torch.no_grad():
                    alpha = v / (v - iou + (1 + eps))
                return iou - (rho2 / c2 + v * alpha)  # CIoU
            return iou - rho2 / c2  # DIoU
        c_area = cw * ch + eps  # convex area
        return iou - (c_area - union) / c_area  # GIoU https://arxiv.org/pdf/1902.09630.pdf
    return iou  # IoU

修改为：

def bbox_iou(box1, box2, xywh=True, GIoU=False, DIoU=False, CIoU=False, EIoU=False, SIoU=False, MPDIoU=False, feat_h=640, feat_w=640,shape_iou=False, scale=0,
             inner_iou=False,  inner_giou=False, inner_diou=False, inner_ciou=False, inner_eiou=False, inner_siou=False, ratio=0.8, eps=1e-7):
    # Returns Intersection over Union (IoU) of box1(1,4) to box2(n,4)

    # Get the coordinates of bounding boxes
    if xywh:  # transform from xywh to xyxy
        (x1, y1, w1, h1), (x2, y2, w2, h2) = box1.chunk(4, -1), box2.chunk(4, -1)
        w1_, h1_, w2_, h2_ = w1 / 2, h1 / 2, w2 / 2, h2 / 2
        b1_x1, b1_x2, b1_y1, b1_y2 = x1 - w1_, x1 + w1_, y1 - h1_, y1 + h1_
        b2_x1, b2_x2, b2_y1, b2_y2 = x2 - w2_, x2 + w2_, y2 - h2_, y2 + h2_
        inner_b1_x1, inner_b1_x2, inner_b1_y1, inner_b1_y2 = x1 - w1_* ratio, x1 + w1_ * ratio, y1 - h1_ * ratio, y1 + h1_ * ratio
        inner_b2_x1, inner_b2_x2, inner_b2_y1, inner_b2_y2 = x2 - w2_* ratio, x2 + w2_ * ratio, y2 - h2_ * ratio, y2 + h2_ * ratio
    else:  # x1, y1, x2, y2 = box1
        b1_x1, b1_y1, b1_x2, b1_y2 = box1.chunk(4, -1)
        b2_x1, b2_y1, b2_x2, b2_y2 = box2.chunk(4, -1)
        w1, h1 = b1_x2 - b1_x1, b1_y2 - b1_y1 + eps
        w2, h2 = b2_x2 - b2_x1, b2_y2 - b2_y1 + eps

    # Intersection area
    inter = (torch.min(b1_x2, b2_x2) - torch.max(b1_x1, b2_x1)).clamp(0) * \
            (torch.min(b1_y2, b2_y2) - torch.max(b1_y1, b2_y1)).clamp(0)


    inner_inter = (torch.min(inner_b1_x2, inner_b2_x2) - torch.max(inner_b1_x1, inner_b2_x1)).clamp(0) * \
                  (torch.min(inner_b1_y2, inner_b2_y2) - torch.max(inner_b1_y1, inner_b2_y1)).clamp(0)

    # Union Area
    union = w1 * h1 + w2 * h2 - inter + eps
    inner_union = w1 * h1 * (ratio**2) + w2 * h2 * (ratio**2) - inner_inter + eps

    # IoU
    iou = inter / union
    iou_inner = inner_inter / inner_union

    if CIoU or DIoU or GIoU:
        cw = torch.max(b1_x2, b2_x2) - torch.min(b1_x1, b2_x1)  # convex (smallest enclosing box) width
        ch = torch.max(b1_y2, b2_y2) - torch.min(b1_y1, b2_y1)  # convex height
        if CIoU or DIoU:  # Distance or Complete IoU https://arxiv.org/abs/1911.08287v1
            c2 = cw ** 2 + ch ** 2 + eps  # convex diagonal squared
            rho2 = ((b2_x1 + b2_x2 - b1_x1 - b1_x2) ** 2 + (b2_y1 + b2_y2 - b1_y1 - b1_y2) ** 2) / 4  # center dist ** 2
            if CIoU:  # https://github.com/Zzh-tju/DIoU-SSD-pytorch/blob/master/utils/box/box_utils.py#L47
                v = (4 / math.pi ** 2) * torch.pow(torch.atan(w2 / h2) - torch.atan(w1 / h1), 2)
                with torch.no_grad():
                    alpha = v / (v - iou + (1 + eps))
                return iou - (rho2 / c2 + v * alpha)  # CIoU
            return iou - rho2 / c2  # DIoU
        c_area = cw * ch + eps  # convex area
        return iou - (c_area - union) / c_area  # GIoU https://arxiv.org/pdf/1902.09630.pdf
    elif EIoU:
        cw = torch.max(b1_x2, b2_x2) - torch.min(b1_x1, b2_x1)  # convex (smallest enclosing box) width
        ch = torch.max(b1_y2, b2_y2) - torch.min(b1_y1, b2_y1)  # convex height
        rho2 = ((b2_x1 + b2_x2 - b1_x1 - b1_x2) ** 2 + (
                    b2_y1 + b2_y2 - b1_y1 - b1_y2) ** 2) / 4  # center distance squared
        c2 = (cw ** 2 + ch ** 2) + eps
        rho_w2 = (w2 - w1) ** 2
        rho_h2 = (h2 - h1) ** 2
        cw2 = cw ** 2 + eps
        ch2 = ch ** 2 + eps
        return iou - (rho2 / c2 + rho_w2 / cw2 + rho_h2 / ch2)
    elif SIoU:
        cw = torch.max(b1_x2, b2_x2) - torch.min(b1_x1, b2_x1)  # convex (smallest enclosing box) width
        ch = torch.max(b1_y2, b2_y2) - torch.min(b1_y1, b2_y1)  # convex height
        s_cw = (b2_x1 + b2_x2 - b1_x1 - b1_x2) * 0.5
        s_ch = (b2_y1 + b2_y2 - b1_y1 - b1_y2) * 0.5
        sigma = torch.pow(s_cw ** 2 + s_ch ** 2, 0.5)
        sin_alpha_1 = torch.abs(s_cw) / sigma
        sin_alpha_2 = torch.abs(s_ch) / sigma
        threshold = pow(2, 0.5) / 2
        sin_alpha = torch.where(sin_alpha_1 > threshold, sin_alpha_2, sin_alpha_1)
        angle_cost = torch.cos(torch.arcsin(sin_alpha) * 2 - math.pi / 2)
        rho_x = (s_cw / cw) ** 2
        rho_y = (s_ch / ch) ** 2
        gamma = angle_cost - 2
        distance_cost = 2 - torch.exp(gamma * rho_x) - torch.exp(gamma * rho_y)
        omiga_w = torch.abs(w1 - w2) / torch.max(w1, w2)
        omiga_h = torch.abs(h1 - h2) / torch.max(h1, h2)
        shape_cost = torch.pow(1 - torch.exp(-1 * omiga_w), 4) + torch.pow(1 - torch.exp(-1 * omiga_h), 4)
        return iou - 0.5 * (distance_cost + shape_cost)
    elif inner_iou:
       return iou_inner
    elif inner_ciou or inner_diou or inner_giou:
        cw = torch.max(b1_x2, b2_x2) - torch.min(b1_x1, b2_x1)  # convex (smallest enclosing box) width
        ch = torch.max(b1_y2, b2_y2) - torch.min(b1_y1, b2_y1)  # convex height
        if CIoU or DIoU:  # Distance or Complete IoU https://arxiv.org/abs/1911.08287v1
            c2 = cw ** 2 + ch ** 2 + eps  # convex diagonal squared
            rho2 = ((b2_x1 + b2_x2 - b1_x1 - b1_x2) ** 2 + (b2_y1 + b2_y2 - b1_y1 - b1_y2) ** 2) / 4  # center dist ** 2
            if CIoU:  # https://github.com/Zzh-tju/DIoU-SSD-pytorch/blob/master/utils/box/box_utils.py#L47
                v = (4 / math.pi ** 2) * torch.pow(torch.atan(w2 / h2) - torch.atan(w1 / h1), 2)
                with torch.no_grad():
                    alpha = v / (v - iou + (1 + eps))
                return iou_inner - (rho2 / c2 + v * alpha)  # CIoU
            return iou_inner - rho2 / c2  # DIoU
        c_area = cw * ch + eps  # convex area
        return iou_inner - (c_area - union) / c_area  # GIoU https://arxiv.org/pdf/1902.09630.pdf
    elif inner_eiou:
        cw = torch.max(b1_x2, b2_x2) - torch.min(b1_x1, b2_x1)  # convex (smallest enclosing box) width
        ch = torch.max(b1_y2, b2_y2) - torch.min(b1_y1, b2_y1)  # convex height
        rho2 = ((b2_x1 + b2_x2 - b1_x1 - b1_x2) ** 2 + (
                    b2_y1 + b2_y2 - b1_y1 - b1_y2) ** 2) / 4  # center distance squared
        c2 = (cw ** 2 + ch ** 2) + eps
        rho_w2 = (w2 - w1) ** 2
        rho_h2 = (h2 - h1) ** 2
        cw2 = cw ** 2 + eps
        ch2 = ch ** 2 + eps
        return iou_inner - (rho2 / c2 + rho_w2 / cw2 + rho_h2 / ch2)
    elif inner_siou:
        cw = torch.max(b1_x2, b2_x2) - torch.min(b1_x1, b2_x1)  # convex (smallest enclosing box) width
        ch = torch.max(b1_y2, b2_y2) - torch.min(b1_y1, b2_y1)  # convex height
        s_cw = (b2_x1 + b2_x2 - b1_x1 - b1_x2) * 0.5
        s_ch = (b2_y1 + b2_y2 - b1_y1 - b1_y2) * 0.5
        sigma = torch.pow(s_cw ** 2 + s_ch ** 2, 0.5)
        sin_alpha_1 = torch.abs(s_cw) / sigma
        sin_alpha_2 = torch.abs(s_ch) / sigma
        threshold = pow(2, 0.5) / 2
        sin_alpha = torch.where(sin_alpha_1 > threshold, sin_alpha_2, sin_alpha_1)
        angle_cost = torch.cos(torch.arcsin(sin_alpha) * 2 - math.pi / 2)
        rho_x = (s_cw / cw) ** 2
        rho_y = (s_ch / ch) ** 2
        gamma = angle_cost - 2
        distance_cost = 2 - torch.exp(gamma * rho_x) - torch.exp(gamma * rho_y)
        omiga_w = torch.abs(w1 - w2) / torch.max(w1, w2)
        omiga_h = torch.abs(h1 - h2) / torch.max(h1, h2)
        shape_cost = torch.pow(1 - torch.exp(-1 * omiga_w), 4) + torch.pow(1 - torch.exp(-1 * omiga_h), 4)
        return iou_inner - 0.5 * (distance_cost + shape_cost)
    elif MPDIoU:
        d1 = (b2_x1 - b1_x1) ** 2 + (b2_y1 - b1_y1) ** 2
        d2 = (b2_x2 - b1_x2) ** 2 + (b2_y2 - b1_y2) ** 2
        mpdiou_hw_pow = feat_h ** 2 + feat_w ** 2
        return iou - d1 / mpdiou_hw_pow - d2 / mpdiou_hw_pow  # MPDIoU
    elif shape_iou:
        ww = 2 * torch.pow(w2, scale) / (torch.pow(w2, scale) + torch.pow(h2, scale))
        hh = 2 * torch.pow(h2, scale) / (torch.pow(w2, scale) + torch.pow(h2, scale))
        cw = torch.max(b1_x2, b2_x2) - torch.min(b1_x1, b2_x1)  # convex width
        ch = torch.max(b1_y2, b2_y2) - torch.min(b1_y1, b2_y1)  # convex height
        c2 = cw ** 2 + ch ** 2 + eps  # convex diagonal squared
        center_distance_x = ((b2_x1 + b2_x2 - b1_x1 - b1_x2) ** 2) / 4
        center_distance_y = ((b2_y1 + b2_y2 - b1_y1 - b1_y2) ** 2) / 4
        center_distance = hh * center_distance_x + ww * center_distance_y
        distance = center_distance / c2

        omiga_w = hh * torch.abs(w1 - w2) / torch.max(w1, w2)
        omiga_h = ww * torch.abs(h1 - h2) / torch.max(h1, h2)
        shape_cost = torch.pow(1 - torch.exp(-1 * omiga_w), 4) + torch.pow(1 - torch.exp(-1 * omiga_h), 4)
        iou = iou - distance - 0.5 * (shape_cost)
    return iou  # IoU

2.使用inner_iou(YOLOv5/YOLOv9)

💯在loss.py中找到class ComputeLoss:

def __call__(self, p, targets):

将：

iou = bbox_iou(pbox, tbox[i], CIoU=True).squeeze()  # iou(prediction, target)

替换为：

iou = bbox_iou(pbox, tbox[i], inner_iou=True).squeeze()  
iou = bbox_iou(pbox, tbox[i], inner_giou=True).squeeze() 
iou = bbox_iou(pbox, tbox[i], inner_diou=True).squeeze() 
iou = bbox_iou(pbox, tbox[i], inner_ciou=True).squeeze() 
iou = bbox_iou(pbox, tbox[i], inner_eiou=True).squeeze() 
iou = bbox_iou(pbox, tbox[i], inner_siou=True).squeeze()

3.使用inner_iou(YOLOv8）

💯在loss.py中找到def _get_loss_bbox(self, pred_bboxes, gt_bboxes, postfix=""):

将：

loss[name_giou] = 1.0 - bbox_iou(pred_bboxes, gt_bboxes, xywh=True, GIoU=True)

替换为：

loss[name_giou] = 1.0 - bbox_iou(pred_bboxes, gt_bboxes, xywh=True, inner_iou=True) 
loss[name_giou] = 1.0 - bbox_iou(pred_bboxes, gt_bboxes, xywh=True, inner_giou=True)
loss[name_giou] = 1.0 - bbox_iou(pred_bboxes, gt_bboxes, xywh=True, inner_diou=True)
loss[name_giou] = 1.0 - bbox_iou(pred_bboxes, gt_bboxes, xywh=True, inner_ciou=True)
loss[name_giou] = 1.0 - bbox_iou(pred_bboxes, gt_bboxes, xywh=True, inner_eiou=True)
loss[name_giou] = 1.0 - bbox_iou(pred_bboxes, gt_bboxes, xywh=True, inner_siou=True)