【深度学习】Focal Loss

https://mp.weixin.qq.com/s/s6Hl8g4mCq-AiMxR03Y6KQ

Focal loss用于解决一阶段目标检测场景中前后背景不平衡的问题。使用此损失让网络学习过程中更加专注于难例。

down-weight easy examples and thus focus training on hard negatives

easily classified (pt ≫ 0.5)

Focal loss

$$\mathrm{FL}\left(p_{\mathrm{t}}\right)=-{\alpha }_{\mathrm{t}}{\left(1-p_{\mathrm{t}}\right)}^{\gamma }\log \left(p_{\mathrm{t}}\right)$$

$$p_{\mathrm{t}}=\{\begin{array}{ll}p& \text{ if }y=1\\ 1-p& \text{ otherwise }\end{array}$$

As pt → 1, the factor goes to 0 and the loss for well-classified examples is down-weighted.

tensorflow实现

import tensorflow as tf

def focal_loss_sigmoid(labels, logits, gamma=2, alpha=0.25):
    """
        Computer focal loss for binary classification
        Args:
          labels: A int32 tensor of shape [batch_size].
          logits: A float32 tensor of shape [batch_size].
          alpha: class weights
          gamma: A scalar for focal loss gamma hyper-parameter.
        Returns:
          A tensor of the same shape as `lables`
        """
    logits = tf.nn.sigmoid(logits)
    focal_loss = -labels * (1 - alpha) * tf.pow(1 - logits, gamma) * tf.log(logits) \
                 - (1 - labels) * (alpha) * tf.pow(logits, gamma) * tf.log(1 - logits)
    return focal_loss

参考：focal_loss