updates

train.py

@@ -227,7 +227,7 @@ def train(
         # Calculate mAP (always test final epoch, skip first 5 if opt.nosave)
         if not (opt.notest or (opt.nosave and epoch < 10)) or epoch == epochs - 1:
             with torch.no_grad():
-                results, maps = test.test(cfg, data_cfg, batch_size=batch_size*2, img_size=img_size_test, model=model,
+                results, maps = test.test(cfg, data_cfg, batch_size=batch_size * 2, img_size=img_size_test, model=model,
                                           conf_thres=0.1)
 
         # Write epoch results

utils/gcp.sh

@@ -93,10 +93,8 @@ python3 test.py --data ../supermarket2/supermarket2.data --weights ../darknet/ba
 
 
 
-
-
 # Debug/Development
-python3 train.py --data data/coco.data --img-size 320 --single-scale --batch-size 64 --accumulate 1 --epochs 1 --evolve
+python3 train.py --data data/coco.data --img-size 320 --single-scale --batch-size 64 --accumulate 1 --epochs 1 --evolve --giou
 python3 test.py --weights weights/latest.pt --cfg cfg/yolov3-spp.cfg --img-size 320
 
 gsutil cp evolve.txt gs://ultralytics

utils/utils.py

@@ -291,7 +291,7 @@ def compute_loss(p, targets, model, giou_loss=False):  # predictions, targets, m
         # Compute losses
         if len(b):  # number of targets
             pi = pi0[b, a, gj, gi]  # predictions closest to anchors
-            tconf[b, a, gj, gi] = 1  # conf
+            tconf[b, a, gj, gi] = 1.0  # conf
             # pi[..., 2:4] = torch.sigmoid(pi[..., 2:4])  # wh power loss (uncomment)
 
             if giou_loss: