HongChuang/ultralytics-ascend
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 3

Backward compatibility support for legacy models (#17010)

Browse source 
Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
This commit is contained in:
Laughing 2024-10-19 02:49:21 +08:00 • committed by GitHub
parent 8d7d1fe390
commit 899fb0495e
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
2 changed files with 18 additions and 8 deletions
Download patch file Download diff file Expand all files Collapse all files

17
ultralytics/nn/modules/head.py
View file

@ -28,6 +28,7 @@ class Detect(nn.Module):
shape = None
anchors = torch.empty(0) # init
strides = torch.empty(0) # init
legacy = False # backward compatibility for v3/v5/v8/v9 models
def __init__(self, nc=80, ch=()):
"""Initializes the YOLO detection layer with specified number of classes and channels."""
@ -41,13 +42,17 @@ class Detect(nn.Module):
self.cv2 = nn.ModuleList(
nn.Sequential(Conv(x, c2, 3), Conv(c2, c2, 3), nn.Conv2d(c2, 4 * self.reg_max, 1)) for x in ch
)
self.cv3 = nn.ModuleList(
nn.Sequential(
nn.Sequential(DWConv(x, x, 3), Conv(x, c3, 1)),
nn.Sequential(DWConv(c3, c3, 3), Conv(c3, c3, 1)),
nn.Conv2d(c3, self.nc, 1),
self.cv3 = (
nn.ModuleList(nn.Sequential(Conv(x, c3, 3), Conv(c3, c3, 3), nn.Conv2d(c3, self.nc, 1)) for x in ch)
if self.legacy
else nn.ModuleList(
nn.Sequential(
nn.Sequential(DWConv(x, x, 3), Conv(x, c3, 1)),
nn.Sequential(DWConv(c3, c3, 3), Conv(c3, c3, 1)),
nn.Conv2d(c3, self.nc, 1),
)
for x in ch
)
for x in ch
)
self.dfl = DFL(self.reg_max) if self.reg_max > 1 else nn.Identity()

9
ultralytics/nn/tasks.py
View file

@ -936,6 +936,7 @@ def parse_model(d, ch, verbose=True): # model_dict, input_channels(3)
import ast
# Args
legacy = True # backward compatibility for v3/v5/v8/v9 models
max_channels = float("inf")
nc, act, scales = (d.get(x) for x in ("nc", "activation", "scales"))
depth, width, kpt_shape = (d.get(x, 1.0) for x in ("depth_multiple", "width_multiple", "kpt_shape"))
@ -1027,8 +1028,10 @@ def parse_model(d, ch, verbose=True): # model_dict, input_channels(3)
}:
args.insert(2, n) # number of repeats
n = 1
if m is C3k2 and scale in "mlx": # for M/L/X sizes
args[3] = True
if m is C3k2: # for M/L/X sizes
legacy = False
if scale in "mlx":
args[3] = True
elif m is AIFI:
args = [ch[f], *args]
elif m in {HGStem, HGBlock}:
@ -1047,6 +1050,8 @@ def parse_model(d, ch, verbose=True): # model_dict, input_channels(3)
args.append([ch[x] for x in f])
if m is Segment:
args[2] = make_divisible(min(args[2], max_channels) * width, 8)
if m in {Detect, Segment, Pose, OBB}:
m.legacy = legacy
elif m is RTDETRDecoder: # special case, channels arg must be passed in index 1
args.insert(1, [ch[x] for x in f])
elif m is CBLinear: