VarifocalLoss在Yolov8中的应用
调用VFL Loss
- 在ultralytics/utils/loss.py可以发现v8实现了VarifocalLoss，但是好像和原论文有点不一样，这里有待考证
- 原文地址：论文
- 在cls损失处
# Cls loss
loss[1] = self.varifocal_loss(pred_scores, target_scores, target_labels) / target_scores_sum # VFL way
# loss[1] = self.bce(pred_scores, target_scores.to(dtype)).sum() / target_scores_sum # BCE
这里可以看到调用varifocal_loss的地方是注释的，同时里面的target_labels是找不到的
实现损失的计算
- 将
_, target_bboxes, target_scores, fg_mask, _ = self.assigner替换为target_labels,…具体如下
target_labels, target_bboxes, target_scores, fg_mask, _ = self.assigner(
pred_scores.detach().sigmoid(),
(pred_bboxes.detach() * stride_tensor).type(gt_bboxes.dtype),
anchor_points * stride_tensor,
gt_labels,
gt_bboxes,
mask_gt,
)
- 本人找到了两种处理target_labels的方法，建议第二种，官方认证github issues
第一种：target_labels = torch.where(target_scores > 0 , 1, 0)
第二种：
target_labels = target_labels.unsqueeze(-1).expand(-1, -1, self.nc) # self.nc: class num
one_hot = torch.zeros(target_labels.size(), device=self.device)
target_labels = one_hot.scatter_(-1, target_labels, 1)
- 完整代码
class v8DetectionLoss:
"""Criterion class for computing training losses."""
def __init__(self, model): # model must be de-paralleled
"""Initializes v8DetectionLoss with the model, defining model-related properties and BCE loss function."""
device = next(model.parameters()).device # get model device
h = model.args # hyperparameters
# import ipdb;ipdb.set_trace()
m = model.model[-1] # Detect() module
self.bce = nn.BCEWithLogitsLoss(reduction="none")
self.hyp = h
self.stride = m.stride # model strides
self.nc = m.nc # number of classes
self.no = m.nc + m.reg_max * 4
self.reg_max = m.reg_max
self.device = device
self.varifocal_loss=VarifocalLoss().to(device)
self.use_dfl = m.reg_max > 1
self.assigner = TaskAlignedAssigner(topk=10, num_classes=self.nc, alpha=0.5, beta=6.0)
self.bbox_loss = BboxLoss(m.reg_max - 1, use_dfl=self.use_dfl).to(device)
self.proj = torch.arange(m.reg_max, dtype=torch.float, device=device)
def preprocess(self, targets, batch_size, scale_tensor):
"""Preprocesses the target counts and matches with the input batch size to output a tensor."""
if targets.shape[0] == 0:
out = torch.zeros(batch_size, 0, 5, device=self.device)
else:
i = targets[:, 0] # image index
_, counts = i.unique(return_counts=True)
counts = counts.to(dtype=torch.int32)
out = torch.zeros(batch_size, counts.max(), 5, device=self.device)
for j in range(batch_size):
matches = i == j
n = matches.sum()
if n:
out[j, :n] = targets[matches, 1:]
out[..., 1:5] = xywh2xyxy(out[..., 1:5].mul_(scale_tensor))
return out
def bbox_decode(self, anchor_points, pred_dist):
"""Decode predicted object bounding box coordinates from anchor points and distribution."""
if self.use_dfl:
b, a, c = pred_dist.shape # batch, anchors, channels
pred_dist = pred_dist.view(b, a, 4, c // 4).softmax(3).matmul(self.proj.type(pred_dist.dtype))
# pred_dist = pred_dist.view(b, a, c // 4, 4).transpose(2,3).softmax(3).matmul(self.proj.type(pred_dist.dtype))
# pred_dist = (pred_dist.view(b, a, c // 4, 4).softmax(2) * self.proj.type(pred_dist.dtype).view(1, 1, -1, 1)).sum(2)
return dist2bbox(pred_dist, anchor_points, xywh=False)
def __call__(self, preds, batch):
"""Calculate the sum of the loss for box, cls and dfl multiplied by batch size."""
loss = torch.zeros(3, device=self.device) # box, cls, dfl
feats = preds[1] if isinstance(preds, tuple) else preds
pred_distri, pred_scores = torch.cat([xi.view(feats[0].shape[0], self.no, -1) for xi in feats], 2).split(
(self.reg_max * 4, self.nc), 1
)
pred_scores = pred_scores.permute(0, 2, 1).contiguous()
pred_distri = pred_distri.permute(0, 2, 1).contiguous()
dtype = pred_scores.dtype
batch_size = pred_scores.shape[0]
imgsz = torch.tensor(feats[0].shape[2:], device=self.device, dtype=dtype) * self.stride[0] # image size (h,w)
anchor_points, stride_tensor = make_anchors(feats, self.stride, 0.5)
# Targets
targets = torch.cat((batch["batch_idx"].view(-1, 1), batch["cls"].view(-1, 1), batch["bboxes"]), 1)
targets = self.preprocess(targets.to(self.device), batch_size, scale_tensor=imgsz[[1, 0, 1, 0]])
gt_labels, gt_bboxes = targets.split((1, 4), 2) # cls, xyxy
mask_gt = gt_bboxes.sum(2, keepdim=True).gt_(0)
# Pboxes
pred_bboxes = self.bbox_decode(anchor_points, pred_distri) # xyxy, (b, h*w, 4)
target_labels, target_bboxes, target_scores, fg_mask, _ = self.assigner(
pred_scores.detach().sigmoid(),
(pred_bboxes.detach() * stride_tensor).type(gt_bboxes.dtype),
anchor_points * stride_tensor,
gt_labels,
gt_bboxes,
mask_gt,
)
target_scores_sum = max(target_scores.sum(), 1)
# target_labels = torch.where(target_scores > 0 , 1, 0)
target_labels = target_labels.unsqueeze(-1).expand(-1, -1, self.nc) # self.nc: class num
one_hot = torch.zeros(target_labels.size(), device=self.device)
target_labels = one_hot.scatter_(-1, target_labels, 1)
# Cls loss
loss[1] = self.varifocal_loss(pred_scores, target_scores, target_labels) / target_scores_sum # VFL way
# loss[1] = self.bce(pred_scores, target_scores.to(dtype)).sum() / target_scores_sum # BCE
# Bbox loss
if fg_mask.sum():
target_bboxes /= stride_tensor
loss[0], loss[2] = self.bbox_loss(
pred_distri, pred_bboxes, anchor_points, target_bboxes, target_scores, target_scores_sum, fg_mask
)
loss[0] *= self.hyp.box # box gain
loss[1] *= self.hyp.cls # cls gain
loss[2] *= self.hyp.dfl # dfl gain
return loss.sum() * batch_size, loss.detach() # loss(box, cls, dfl)
参考
参考1
参考2
参考3