Learn about build targets

[WZMIAOMIAO]

Search before asking

• I have searched the YOLOv5 issues and discussions and found no similar questions.

Question

First of all, thank you for your great works.

I read and learn the code:

yolov5/utils/loss.py

Lines 169 to 222 in 8a66eba

	def build_targets(self, p, targets):
	# Build targets for compute_loss(), input targets(image,class,x,y,w,h)
	na, nt = self.na, targets.shape[0] # number of anchors, targets
	tcls, tbox, indices, anch = [], [], [], []
	gain = torch.ones(7, device=targets.device) # normalized to gridspace gain
	ai = torch.arange(na, device=targets.device).float().view(na, 1).repeat(1, nt) # same as .repeat_interleave(nt)
	targets = torch.cat((targets.repeat(na, 1, 1), ai[:, :, None]), 2) # append anchor indices
	g = 0.5 # bias
	off = torch.tensor([[0, 0],
	[1, 0], [0, 1], [-1, 0], [0, -1], # j,k,l,m
	# [1, 1], [1, -1], [-1, 1], [-1, -1], # jk,jm,lk,lm
	], device=targets.device).float() * g # offsets
	for i in range(self.nl):
	anchors = self.anchors[i]
	gain[2:6] = torch.tensor(p[i].shape)[[3, 2, 3, 2]] # xyxy gain
	# Match targets to anchors
	t = targets * gain
	if nt:
	# Matches
	r = t[:, :, 4:6] / anchors[:, None] # wh ratio
	j = torch.max(r, 1 / r).max(2)[0] < self.hyp['anchor_t'] # compare
	# j = wh_iou(anchors, t[:, 4:6]) > model.hyp['iou_t'] # iou(3,n)=wh_iou(anchors(3,2), gwh(n,2))
	t = t[j] # filter
	# Offsets
	gxy = t[:, 2:4] # grid xy
	gxi = gain[[2, 3]] - gxy # inverse
	j, k = ((gxy % 1 < g) & (gxy > 1)).T
	l, m = ((gxi % 1 < g) & (gxi > 1)).T
	j = torch.stack((torch.ones_like(j), j, k, l, m))
	t = t.repeat((5, 1, 1))[j]
	offsets = (torch.zeros_like(gxy)[None] + off[:, None])[j]
	else:
	t = targets[0]
	offsets = 0
	# Define
	b, c = t[:, :2].long().T # image, class
	gxy = t[:, 2:4] # grid xy
	gwh = t[:, 4:6] # grid wh
	gij = (gxy - offsets).long()
	gi, gj = gij.T # grid xy indices
	# Append
	a = t[:, 6].long() # anchor indices
	indices.append((b, a, gj.clamp_(0, gain[3] - 1), gi.clamp_(0, gain[2] - 1))) # image, anchor, grid indices
	tbox.append(torch.cat((gxy - gij, gwh), 1)) # box
	anch.append(anchors[a]) # anchors
	tcls.append(c) # class
	return tcls, tbox, indices, anch

My guess is that offsets are introduced to get more positive samples. Then, I want to know whether I understand correctly. Here is the diagram I drew:

If I don't understand correctly, I hope you can give me some tips. Thank you!

Additional

No response

[glenn-jocher]

@WZMIAOMIAO yes this is correct!

[WZMIAOMIAO]

@WZMIAOMIAO yes this is correct!

thank you for your reply.

[WZMIAOMIAO]

#6998

[myasser63]

@WZMIAOMIAO Can you explain this more in your summary?