Focal Loss是一种针对类别不平衡的损失函数，可以在训练过程中减少易分类样本的权重，从而提高模型对难分类样本的关注度。以下是将SSD的损失函数改成focal loss的代码：
```python
import torch
import torch.nn as nn
class FocalLoss(nn.Module):
def __init__(self, alpha=0.25, gamma=2, logits=True, reduction='mean'):
super(FocalLoss, self).__init__()
self.alpha = alpha
self.gamma = gamma
self.logits = logits
self.reduction = reduction
def forward(self, inputs, targets):
if self.logits:
BCE_loss = nn.functional.binary_cross_entropy_with_logits(inputs, targets, reduction='none')
else:
BCE_loss = nn.functional.binary_cross_entropy(inputs, targets, reduction='none')
pt = torch.exp(-BCE_loss)
F_loss = self.alpha * (1 - pt) ** self.gamma * BCE_loss
if self.reduction == 'mean':
return torch.mean(F_loss)
elif self.reduction == 'sum':
return torch.sum(F_loss)
else:
return F_loss
class MultiBoxLoss(nn.Module):
def __init__(self, num_classes, overlap_thresh, prior_for_matching,
bkg_label, neg_mining, neg_pos, neg_overlap, encode_target,
use_gpu=True):
super(MultiBoxLoss, self).__init__()
self.use_gpu = use_gpu
self.num_classes = num_classes
self.threshold = overlap_thresh
self.background_label = bkg_label
self.encode_target = encode_target
self.use_prior_for_matching = prior_for_matching
self.do_neg_mining = neg_mining
self.negpos_ratio = neg_pos
self.neg_overlap = neg_overlap
self.variance = [0.1, 0.2]
self.focal_loss = FocalLoss()
def forward(self, predictions, targets):
loc_data, conf_data, prior_data = predictions
num = loc_data.size(0)
num_priors = prior_data.size(0)
loc_t = torch.Tensor(num, num_priors, 4)
conf_t = torch.LongTensor(num, num_priors)
for idx in range(num):
truths = targets[idx][:, :-1].data
labels = targets[idx][:, -1].data
defaults = prior_data.data
match(self.threshold, truths, defaults, self.variance, labels,
loc_t, conf_t, idx)
if self.use_gpu:
loc_t = loc_t.cuda()
conf_t = conf_t.cuda()
pos = conf_t > 0
num_pos = pos.sum(dim=1, keepdim=True)
# Localization Loss (Smooth L1)
# Shape: [batch,num_priors,4]
pos_idx = pos.unsqueeze(pos.dim()).expand_as(loc_data)
loc_p = loc_data[pos_idx].view(-1, 4)
loc_t = loc_t[pos_idx].view(-1, 4)
loss_l = nn.functional.smooth_l1_loss(loc_p, loc_t, reduction='sum')
# Compute max conf across batch for hard negative mining
batch_conf = conf_data.view(-1, self.num_classes)
loss_c = self.focal_loss(batch_conf, conf_t.view(-1, 1))
# Hard Negative Mining
loss_c[pos] = 0 # filter out pos boxes for now
loss_c = loss_c.view(num, -1)
_, loss_idx = loss_c.sort(1, descending=True)
_, idx_rank = loss_idx.sort(1)
num_pos = pos.long().sum(1, keepdim=True)
num_neg = torch.clamp(self.negpos_ratio * num_pos, max=pos.size(1) - 1)
neg = idx_rank < num_neg.expand_as(idx_rank)
# Confidence Loss Including Positive and Negative Examples
pos_idx = pos.unsqueeze(2).expand_as(conf_data)
neg_idx = neg.unsqueeze(2).expand_as(conf_data)
conf_p = conf_data[(pos_idx + neg_idx).gt(0)].view(-1, self.num_classes)
targets_weighted = conf_t[(pos + neg).gt(0)]
loss_c = self.focal_loss(conf_p, targets_weighted)
# Sum of losses: L(x,c,l,g) = (Lconf(x, c) + αLloc(x,l,g)) / N
N = num_pos.sum().float()
loss_l /= N
loss_c /= N
return loss_l, loss_c
```
在MultiBoxLoss中，我们用focal_loss替换了原来的交叉熵损失函数。在FocalLoss中，我们计算每个样本的二元交叉熵损失，然后再乘以一个类别权重系数(1 - pt)^gamma，其中pt是预测概率的指数形式，gamma是一个可调参数，用于控制易分类样本的权重。最后，我们返回一个平均的损失值。在MultiBoxLoss中，我们计算了定位损失和分类损失，并将它们相加，再除以正样本的数量求取平均值。同时，我们采用了硬负样本挖掘策略，过滤掉难以分类的样本，提高模型的准确率。
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
1万+
