Update utils.py

jsago · Mar 29, 2020 · 8108f41 · 8108f41
1 parent 6c19620
commit 8108f41
Showing 1 changed file with 8 additions and 6 deletions.
diff --git a/utils/utils.py b/utils/utils.py
@@ -424,6 +424,7 @@ def smooth_BCE(eps=0.1):
     return 1.0 - 0.5 * eps, 0.5 * eps
 
 def compute_loss(p, targets, model):
+    # p: (bs, anchors, grid, grid, classes + xywh)
     # predictions, targets, model
     ft = torch.cuda.FloatTensor if p[0].is_cuda else torch.Tensor
     lcls, lbox, lobj = ft([0]), ft([0]), ft([0])
@@ -511,14 +512,14 @@ def compute_loss(p, targets, model):
         elif 'BCE' in arc:  # unified BCE (80 classes)
             t = torch.zeros_like(pi[..., 5:])  # targets
             if nb:
-                t[b, a, gj, gi, tcls[i]] = 1.0 # 对应class置信度设置为1
+                t[b, a, gj, gi, tcls[i]] = 1.0 # 对应正样本class置信度设置为1
             lobj += BCE(pi[..., 5:], t)
             #pi[...,5:]对应的是所有的class
 
         elif 'CE' in arc:  # unified CE (1 background + 80 classes)
             t = torch.zeros_like(pi[..., 0], dtype=torch.long)  # targets
             if nb:
-                t[b, a, gj, gi] = tcls[i] + 1
+                t[b, a, gj, gi] = tcls[i] + 1 # 由于cls是从零开始计数的，所以+1
             lcls += CE(pi[..., 4:].view(-1, model.nc + 1), t.view(-1))
             # 这里将obj loss和cls loss一起计算，使用CrossEntropy Loss
 
@@ -529,9 +530,11 @@ def compute_loss(p, targets, model):
 
     if red == 'sum':
         bs = tobj.shape[0]  # batch size
-        lobj *= 2 / (2535 * bs) * 2  # 3 / np * 2
-        # from 6300 to (13**2+26**2)*3 = 2535
-        # 3代表3个yolo层
+        lobj *= 3 / (6300 * bs) * 2
+        # 6300 = (10 ** 2 + 20 ** 2 + 40 ** 2) * 3
+        # 输入为320x320的图片，则存在6300个anchor
+        # 3代表3个yolo层, 2是一个超参数，通过实验获取
+        # 如果不想计算的话，可以修改red='mean'
         if ng:
             lcls *= 3 / ng / model.nc
             lbox *= 3 / ng
@@ -633,7 +636,6 @@ def build_targets(model, targets):
             assert c.max() < model.nc, 'Model accepts %g classes labeled from 0-%g, however you labelled a class %g. ' \
                                        'See https://github.com/ultralytics/yolov3/wiki/Train-Custom-Data' % (
                                            model.nc, model.nc - 1, c.max())
-    # tcls, tbox, indices, anchor_vec = build_targets(model, targets)
     return tcls, tbox, indices, av