Bienqq
/
car-detection-bayes
2
1
Fork 0
Code Issues Pull Requests Releases Wiki Activity

updates

This commit is contained in:
Glenn Jocher 2019-08-18 02:02:04 +02:00
parent 0aece25ef6
commit 43230c48bf
2 changed files with 39 additions and 38 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

74
train.py
View File

@ -33,45 +33,45 @@ except:
# 0.268 0.268 0.178 0.240 4.36 1.104 5.596 0.2087 14.47 2.599 16.27 2.406 0.4114 0.001585 -4 0.950 0.000524 # hc
# 0.161 0.327 0.190 0.193 7.82 1.153 4.062 0.1845 24.28 3.05 20.93 2.842 0.2759 0.001357 -4 0.916 0.000572 # hd 0.438 mAP @ epoch 100
# Hyperparameters (j-series, 50.5 mAP yolov3-320) evolved by @ktian08 https://github.com/ultralytics/yolov3/issues/310
hyp = {'giou': 1.582, # giou loss gain
'xy': 4.688, # xy loss gain
'wh': 0.1857, # wh loss gain
'cls': 27.76, # cls loss gain (CE=~1.0, uCE=~20, uBCE=~200,~30)
'cls_pw': 1.446, # cls BCELoss positive_weight
'obj': 21.35, # obj loss gain
'obj_pw': 3.941, # obj BCELoss positive_weight
'iou_t': 0.2635, # iou training threshold
'lr0': 0.002324, # initial learning rate
'lrf': -4., # final LambdaLR learning rate = lr0 * (10 ** lrf)
'momentum': 0.97, # SGD momentum
'weight_decay': 0.0004569, # optimizer weight decay
'hsv_s': 0.5703, # image HSV-Saturation augmentation (fraction)
'hsv_v': 0.3174, # image HSV-Value augmentation (fraction)
'degrees': 1.113, # image rotation (+/- deg)
'translate': 0.06797, # image translation (+/- fraction)
'scale': 0.1059, # image scale (+/- gain)
'shear': 0.5768} # image shear (+/- deg)
# # Hyperparameters (i-series)
# hyp = {'giou': 1.43, # giou loss gain
# # Hyperparameters (j-series, 50.5 mAP yolov3-320) evolved by @ktian08 https://github.com/ultralytics/yolov3/issues/310
# hyp = {'giou': 1.582, # giou loss gain
# 'xy': 4.688, # xy loss gain
# 'wh': 0.1857, # wh loss gain
# 'cls': 11.7, # cls loss gain
# 'cls_pw': 4.81, # cls BCELoss positive_weight
# 'obj': 11.5, # obj loss gain
# 'obj_pw': 1.56, # obj BCELoss positive_weight
# 'iou_t': 0.281, # iou training threshold
# 'lr0': 0.0013, # initial learning rate
# 'cls': 27.76, # cls loss gain (CE=~1.0, uCE=~20, uBCE=~200,~30)
# 'cls_pw': 1.446, # cls BCELoss positive_weight
# 'obj': 21.35, # obj loss gain
# 'obj_pw': 3.941, # obj BCELoss positive_weight
# 'iou_t': 0.2635, # iou training threshold
# 'lr0': 0.010324, # initial learning rate
# 'lrf': -4., # final LambdaLR learning rate = lr0 * (10 ** lrf)
# 'momentum': 0.944, # SGD momentum
# 'weight_decay': 0.000427, # optimizer weight decay
# 'hsv_s': 0.0599, # image HSV-Saturation augmentation (fraction)
# 'hsv_v': 0.142, # image HSV-Value augmentation (fraction)
# 'degrees': 1.03, # image rotation (+/- deg)
# 'translate': 0.0552, # image translation (+/- fraction)
# 'scale': 0.0555, # image scale (+/- gain)
# 'shear': 0.434} # image shear (+/- deg)
# 'momentum': 0.97, # SGD momentum
# 'weight_decay': 0.0004569, # optimizer weight decay
# 'hsv_s': 0.5703, # image HSV-Saturation augmentation (fraction)
# 'hsv_v': 0.3174, # image HSV-Value augmentation (fraction)
# 'degrees': 1.113, # image rotation (+/- deg)
# 'translate': 0.06797, # image translation (+/- fraction)
# 'scale': 0.1059, # image scale (+/- gain)
# 'shear': 0.5768} # image shear (+/- deg)
# Hyperparameters (i-series)
hyp = {'giou': 1.43, # giou loss gain
'xy': 4.688, # xy loss gain
'wh': 0.1857, # wh loss gain
'cls': 11.7, # cls loss gain
'cls_pw': 4.81, # cls BCELoss positive_weight
'obj': 11.5, # obj loss gain
'obj_pw': 1.56, # obj BCELoss positive_weight
'iou_t': 0.281, # iou training threshold
'lr0': 0.01, # initial learning rate
'lrf': -4., # final LambdaLR learning rate = lr0 * (10 ** lrf)
'momentum': 0.97, # SGD momentum
'weight_decay': 0.000427, # optimizer weight decay
'hsv_s': 0.0599, # image HSV-Saturation augmentation (fraction)
'hsv_v': 0.142, # image HSV-Value augmentation (fraction)
'degrees': 1.03, # image rotation (+/- deg)
'translate': 0.0552, # image translation (+/- fraction)
'scale': 0.0555, # image scale (+/- gain)
'shear': 0.434} # image shear (+/- deg)
def train(cfg,
@ -360,7 +360,7 @@ if __name__ == '__main__':
parser.add_argument('--batch-size', type=int, default=32, help='batch size')
parser.add_argument('--accumulate', type=int, default=2, help='number of batches to accumulate before optimizing')
parser.add_argument('--cfg', type=str, default='cfg/yolov3-spp.cfg', help='cfg file path')
parser.add_argument('--data', type=str, default='data/coco.data', help='coco.data file path')
parser.add_argument('--data', type=str, default='data/coco_64img.data', help='coco.data file path')
parser.add_argument('--multi-scale', action='store_true', help='train at (1/1.5)x - 1.5x sizes')
parser.add_argument('--img-size', type=int, default=416, help='inference size (pixels)')
parser.add_argument('--rect', action='store_true', help='rectangular training')

3
utils/utils.py
View File

@ -293,13 +293,14 @@ class FocalLoss(nn.Module):
# i.e. criteria = FocalLoss(nn.BCEWithLogitsLoss(), gamma=2.5)
def __init__(self, loss_fcn, alpha=1, gamma=2, reduction='mean'):
super(FocalLoss, self).__init__()
loss_fcn.reduction = 'none' # required to apply FL to each element
self.loss_fcn = loss_fcn
self.alpha = alpha
self.gamma = gamma
self.reduction = reduction
def forward(self, input, target):
loss = self.loss_fcn(input, target, reduction='none')
loss = self.loss_fcn(input, target)
pt = torch.exp(-loss)
loss *= self.alpha * (1 - pt) ** self.gamma