Changed Optimizer to Alpha ultralytics#10

data/hyps/hyp.scratch-high.yaml

@@ -3,7 +3,7 @@
 # python train.py --batch 32 --cfg yolov5m6.yaml --weights '' --data coco.yaml --img 1280 --epochs 300
 # See tutorials for hyperparameter evolution https://github.com/ultralytics/yolov5#tutorials
 
-lr0: 0.01  # initial learning rate (SGD=1E-2, Adam=1E-3)
+lr0: 0.003  # initial learning rate (SGD=1E-2, Adam=1E-3)
 lrf: 0.1  # final OneCycleLR learning rate (lr0 * lrf)
 momentum: 0.937  # SGD momentum/Adam beta1
 weight_decay: 0.0005  # optimizer weight decay 5e-4

train.py

@@ -489,7 +489,7 @@ def parse_opt(known=False):
     parser.add_argument('--device', default='', help='cuda device, i.e. 0 or 0,1,2,3 or cpu')
     parser.add_argument('--multi-scale', action='store_true', help='vary img-size +/- 50%%')
     parser.add_argument('--single-cls', action='store_true', help='train multi-class data as single-class')
-    parser.add_argument('--optimizer', type=str, choices=['SGD', 'Adam', 'AdamW'], default='SGD', help='optimizer')
+    parser.add_argument('--optimizer', type=str, choices=['SGD', 'Adam', 'AdamW'], default='Adam', help='optimizer')
     parser.add_argument('--sync-bn', action='store_true', help='use SyncBatchNorm, only available in DDP mode')
     parser.add_argument('--workers', type=int, default=8, help='max dataloader workers (per RANK in DDP mode)')
     parser.add_argument('--project', default=ROOT / 'runs/train', help='save to project/name')
@@ -514,7 +514,7 @@ def parse_opt(known=False):
 
 
 def main(opt, callbacks=Callbacks()):
-    print('\n---------- VERSION:', '#0009', '----------\n')
+    print('\n---------- VERSION:', '#0010', '----------\n')
     # Checks
     if RANK in {-1, 0}:
         print_args(vars(opt))

utils/loss.py

@@ -87,7 +87,7 @@ def forward(self, pred, true):
         else:  # 'none'
             return loss
 
-def weighted_bce(y_pred, y_true, BETA=2):
+def weighted_bce(y_pred, y_true, BETA=40):
     weights = (y_true * (BETA - 1)) + 1
     bce = nn.BCEWithLogitsLoss(reduction='none')(y_pred, y_true)
     wbce = torch.mean(bce * weights)

val.py

@@ -321,7 +321,7 @@ def run(
     # Print results per class
     if (verbose or (nc < 50 and not training)) and nc > 1 and len(stats):
         for i, c in enumerate(ap_class):
-            LOGGER.info(pf % (names[c], seen, nt[c], p[i], r[i], ap50[i], ap[i], iou))
+            LOGGER.info(pf % (names[c], seen, nt[c], p[i], r[i], ap50[i], ap[i], iou, rail_iou))
 
     # Print speeds
     t = tuple(x.t / seen * 1E3 for x in dt)  # speeds per image