fix_DPP_mAP_drop

models/common.py

@@ -36,11 +36,14 @@ def autopad(k, p=None):  # kernel, padding
 
 class Conv(nn.Module):
     # Standard convolution
+    default_act = nn.SiLU()  # default activation
+
     def __init__(self, c1, c2, k=1, s=1, p=None, g=1, act=True):  # ch_in, ch_out, kernel, stride, padding, groups
         super().__init__()
         self.conv = nn.Conv2d(c1, c2, k, s, autopad(k, p), groups=g, bias=False)
         self.bn = nn.BatchNorm2d(c2)
-        self.act = nn.SiLU() if act is True else (act if isinstance(act, nn.Module) else nn.Identity())
+        self.act = self.default_act if act is True else act if isinstance(act, nn.Module) else nn.Identity()
+
         self.is_fuse = False
 
     def forward(self, x):

models/yolo.py

@@ -21,7 +21,7 @@
 from models.common import C3, C3SPP, C3TR, SPP, SPPF, Bottleneck, BottleneckCSP, C3Ghost, C3x, Concat, Contract, Conv, CrossConv, DWConv, DWConvTranspose2d, Expand, Focus, GhostBottleneck, GhostConv
 from models.experimental import MixConv2d
 from utils.autoanchor import check_anchor_order
-from utils.general import LOGGER, check_yaml, make_divisible, print_args
+from utils.general import LOGGER, check_yaml, colorstr, make_divisible, print_args
 from utils.oneflow_utils import fuse_conv_and_bn, initialize_weights, model_info, profile, scale_img, select_device, time_sync
 from utils.plots import feature_visualization
 
@@ -271,7 +271,10 @@ def _apply(self, fn):
 
 def parse_model(d, ch):  # model_dict, input_channels(3)
     LOGGER.info(f"\n{'':>3}{'from':>18}{'n':>3}{'params':>10}  {'module':<40}{'arguments':<30}")
-    anchors, nc, gd, gw = d["anchors"], d["nc"], d["depth_multiple"], d["width_multiple"]
+    anchors, nc, gd, gw, act = d["anchors"], d["nc"], d["depth_multiple"], d["width_multiple"], d.get("activation")
+    if act:
+        Conv.default_act = eval(act)  # redefine default activation, i.e. Conv.default_act = nn.SiLU()
+        LOGGER.info(f"{colorstr('activation:')} {act}")  # print
     na = (len(anchors[0]) // 2) if isinstance(anchors, list) else anchors  # number of anchors
     no = na * (nc + 5)  # number of outputs = anchors * (classes + 5)
 

train.py

@@ -125,7 +125,7 @@ def train(hyp, opt, device, callbacks):  # hyp is path/to/hyp.yaml or hyp dictio
     plots = not evolve and not opt.noplots  # create plots
     cuda = device.type != "cpu"
 
-    init_seeds(1, deterministic=True)
+    init_seeds(opt.seed + 1 + RANK, deterministic=True)
 
     # with torch_distributed_zero_first(LOCAL_RANK): # 这个是上下文管理器
     data_dict = data_dict or check_dataset(data)  # check if None

utils/dataloaders.py

@@ -76,7 +76,7 @@
 )  # include video suffixes
 BAR_FORMAT = "{l_bar}{bar:10}{r_bar}{bar:-10b}"  # tqdm bar format
 LOCAL_RANK = int(os.getenv("LOCAL_RANK", -1))  # https://pytorch.org/docs/stable/elastic/run.html
-
+RANK = int(os.getenv("RANK", -1))
 # Get orientation exif tag
 for orientation in ExifTags.TAGS.keys():
     if ExifTags.TAGS[orientation] == "Orientation":
@@ -178,7 +178,7 @@ def create_dataloader(
     sampler = None if rank == -1 else distributed.DistributedSampler(dataset, shuffle=shuffle)
     loader = DataLoader if image_weights else InfiniteDataLoader  # only DataLoader allows for attribute updates
     generator = flow.Generator()
-    generator.manual_seed(0)
+    generator.manual_seed(6148914691236517205 + RANK)
     return (
         loader(
             dataset,