update code

pytorch_object_detection/yolov3_spp/train.py

@@ -174,8 +174,8 @@ def train(hyp):
     # start training
     # caching val_data when you have plenty of memory(RAM)
     print("caching val_data for evaluation.")
-    coco = None
-    # coco = get_coco_api_from_dataset(val_dataset)
+    # coco = None
+    coco = get_coco_api_from_dataset(val_dataset)
     print("starting traning for %g epochs..." % epochs)
     print('Using %g dataloader workers' % nw)
     for epoch in range(start_epoch, epochs):

pytorch_object_detection/yolov3_spp/train_utils/train_eval_utils.py

@@ -80,7 +80,7 @@ def train_one_epoch(model, optimizer, data_loader, device, epoch,
                 print("training image path: {}".format(",".join(paths)))
                 sys.exit(1)
 
-            losses *= 1 / accumulate  # scale loss
+            losses *= 1. / accumulate  # scale loss
 
         # backward
         scaler.scale(losses).backward()

pytorch_object_detection/yolov3_spp/utils/utils.py

@@ -435,18 +435,25 @@ def build_targets(p, targets, model):
 
     style = None
     multi_gpu = type(model) in (nn.parallel.DataParallel, nn.parallel.DistributedDataParallel)
-    for i, j in enumerate(model.yolo_layers):
+    for i, j in enumerate(model.yolo_layers):  # [89, 101, 113]
+        # 获取该yolo predictor对应的anchors
         anchors = model.module.module_list[j].anchor_vec if multi_gpu else model.module_list[j].anchor_vec
         gain[2:] = torch.tensor(p[i].shape)[[3, 2, 3, 2]]  # xyxy gain
         na = anchors.shape[0]  # number of anchors
+        # [3] -> [3, 1] -> [3, nt]
         at = torch.arange(na).view(na, 1).repeat(1, nt)  # anchor tensor, same as .repeat_interleave(nt)
 
         # Match targets to anchors
         a, t, offsets = [], targets * gain, 0
-        if nt:
+        if nt:  # 如果存在target的话
             # r = t[None, :, 4:6] / anchors[:, None]  # wh ratio
             # j = torch.max(r, 1. / r).max(2)[0] < model.hyp['anchor_t']  # compare
+
+            # iou_t = 0.20
+            # j: [3, nt]
             j = wh_iou(anchors, t[:, 4:6]) > model.hyp['iou_t']  # iou(3,n) = wh_iou(anchors(3,2), gwh(n,2))
+            # t.repeat(na, 1, 1): [nt, 6] -> [3, nt, 6]
+            # 获取iou大于阈值的anchor与target对应信息
             a, t = at[j], t.repeat(na, 1, 1)[j]  # filter
 
             # overlaps
@@ -466,18 +473,20 @@ def build_targets(p, targets, model):
                 offsets = torch.cat((z, z[j] + off[0], z[k] + off[1], z[l] + off[2], z[m] + off[3]), 0) * g
 
         # Define
+        # long等于to(torch.int64), 数值向下取整
         b, c = t[:, :2].long().T  # image, class
         gxy = t[:, 2:4]  # grid xy
         gwh = t[:, 4:6]  # grid wh
-        gij = (gxy - offsets).long()
+        gij = (gxy - offsets).long()  # 匹配targets所在的grid cell左上角坐标
         gi, gj = gij.T  # grid xy indices
 
         # Append
-        indices.append((b, a, gj, gi))  # image, anchor, grid indices
-        tbox.append(torch.cat((gxy - gij, gwh), 1))  # box
+        indices.append((b, a, gj, gi))  # image, anchor, grid indices(x, y)
+        tbox.append(torch.cat((gxy - gij, gwh), 1))  # gt box相对anchor的x,y偏移量以及w,h
         anch.append(anchors[a])  # anchors
         tcls.append(c)  # class
         if c.shape[0]:  # if any targets
+            # 目标的标签数值不能大于给定的目标类别数
             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())