RuntimeWarning: invalid value encountered in ulong_scalars

[zhengjie6]

When trying to run python eval.py configs/rn18_single_scale.py, i ran into the follow error in evaluate.py, line 17 and 30

(swiftnet) fyp@fyp-P95-96-97Ex-Rx:~/Desktop/swiftnet$ python eval.py configs/rn18_single_scale.py Num images: 0 Num images: 0 Upsample layer: in = 128, skip = 64, out = 128 /home/fyp/Desktop/swiftnet/models/util.py:44: UserWarning: Using conv type 1x1: <class 'torch.nn.modules.conv.Conv2d'> warnings.warn(f'Using conv type {k}x{k}: {conv_class}') /home/fyp/Desktop/swiftnet/models/util.py:44: UserWarning: Using conv type 3x3: <class 'torch.nn.modules.conv.Conv2d'> warnings.warn(f'Using conv type {k}x{k}: {conv_class}') /home/fyp/Desktop/swiftnet/models/util.py:62: UserWarning: Using skips: True (only skips: False) warnings.warn(f'\tUsing skips: {self.use_skip} (only skips: {self.only_skip})', UserWarning) Upsample layer: in = 128, skip = 128, out = 128 Upsample layer: in = 128, skip = 256, out = 128 Batch size: 14 Num params: 11,797,071 = 620,559(random init) + 11,176,512(fine tune) SPP params: 116,476 0it [00:00, ?it/s]

/home/fyp/Desktop/swiftnet/evaluation/evaluate.py:17: RuntimeWarning: invalid value encountered in ulong_scalars avg_pixel_acc = num_correct / total_size 100.0 > Errors: /home/fyp/Desktop/swiftnet/evaluation/evaluate.py:30: RuntimeWarning: invalid value encountered in ulong_scalars class_iou[i] = (TP / (TP + FP[i] + FN[i])) 100.0 road IoU accuracy = nan % sidewalk IoU accuracy = nan % building IoU accuracy = nan % wall IoU accuracy = nan % fence IoU accuracy = nan % pole IoU accuracy = nan % traffic light IoU accuracy = nan % traffic sign IoU accuracy = nan % vegetation IoU accuracy = nan % terrain IoU accuracy = nan % sky IoU accuracy = nan % person IoU accuracy = nan % rider IoU accuracy = nan % car IoU accuracy = nan % truck IoU accuracy = nan % bus IoU accuracy = nan % train IoU accuracy = nan % motorcycle IoU accuracy = nan % bicycle IoU accuracy = nan % IoU mean class accuracy -> TP / (TP+FN+FP) = nan % mean class recall -> TP / (TP+FN) = 0.00 % mean class precision -> TP / (TP+FP) = 0.00 % pixel accuracy = nan % val: nan 0it [00:00, ?it/s]

Errors: road IoU accuracy = nan % sidewalk IoU accuracy = nan % building IoU accuracy = nan % wall IoU accuracy = nan % fence IoU accuracy = nan % pole IoU accuracy = nan % traffic light IoU accuracy = nan % traffic sign IoU accuracy = nan % vegetation IoU accuracy = nan % terrain IoU accuracy = nan % sky IoU accuracy = nan % person IoU accuracy = nan % rider IoU accuracy = nan % car IoU accuracy = nan % truck IoU accuracy = nan % bus IoU accuracy = nan % train IoU accuracy = nan % motorcycle IoU accuracy = nan % bicycle IoU accuracy = nan % IoU mean class accuracy -> TP / (TP+FN+FP) = nan % mean class recall -> TP / (TP+FN) = 0.00 % mean class precision -> TP / (TP+FP) = 0.00 % pixel accuracy = nan % train: nan