updates

train.py

@@ -197,6 +197,7 @@ def train(cfg,
                                              collate_fn=dataset.collate_fn)
 
     # Start training
+    model.nc = nc  # attach number of classes to model
     model.hyp = hyp  # attach hyperparameters to model
     if dataset.image_weights:
         model.class_weights = labels_to_class_weights(dataset.labels, nc).to(device)  # attach class weights

utils/utils.py

@@ -315,10 +315,11 @@ def compute_loss(p, targets, model, giou_loss=True):  # predictions, targets, mo
                 lxy += (k * h['xy']) * MSE(pxy, txy[i])  # xy loss
                 lwh += (k * h['wh']) * MSE(pi[..., 2:4], twh[i])  # wh yolo loss
 
+            if model.nc > 1:  # cls loss (only if multiple classes)
                 tclsm = torch.zeros_like(pi[..., 5:])
                 tclsm[range(nb), tcls[i]] = 1.0
-            lcls += (k * h['cls']) * BCEcls(pi[..., 5:], tclsm)  # cls loss (BCE)
-            # lcls += (k * h['cls']) * CE(pi[..., 5:], tcls[i])  # cls loss (CE)
+                lcls += (k * h['cls']) * BCEcls(pi[..., 5:], tclsm)  # BCE
+                # lcls += (k * h['cls']) * CE(pi[..., 5:], tcls[i])  # CE
 
             # Append targets to text file
             # with open('targets.txt', 'a') as file:
@@ -383,7 +384,7 @@ def build_targets(model, targets):
 
         # Class
         tcls.append(c)
-        if c.shape[0]:
+        if c.shape[0]:  # if any targets
             assert c.max() <= layer.nc, 'Target classes exceed model classes'
 
     return txy, twh, tcls, tbox, indices, anchor_vec
@@ -777,7 +778,7 @@ def plot_results_overlay(start=1, stop=0):  # from utils.utils import *; plot_re
         fig, ax = plt.subplots(1, 5, figsize=(14, 3.5))
         ax = ax.ravel()
         for i in range(5):
-            for j in [i, i+5]:
+            for j in [i, i + 5]:
                 ax[i].plot(x, results[j, x], marker='.', label=s[j])
             ax[i].set_title(t[i])
             ax[i].legend()