2702 - Focal Loss

参考文章：
https://blog.csdn.net/weixin_44791964/article/details/102853782
参考讲解视频：
https://www.bilibili.com/video/BV1E7411J72R?from=search&seid=3954447457476218575&spm_id_from=333.337.0.0
什么是Focal loss
Focal loss是何恺明大神提出的一种新的loss计算方案。
其具有两个重要的特点。

1、控制正负样本的权重

2、控制容易分类和难分类样本的权重

正负样本的概念如下：
一张图像可能生成成千上万的候选框，但是其中只有很少一部分是包含目标的的，有目标的就是正样本，没有目标的就是负样本。

容易分类和难分类样本的概念如下：
假设存在一个二分类，样本1属于类别1的pt=0.9，样本2属于类别1的pt=0.6，显然前者更可能是类别1，其就是容易分类的样本；后者有可能是类别1，所以其为难分类样本。

def focal(alpha=0.25, gamma=2.0):
    def _focal(y_true, y_pred):
        # y_true [batch_size, num_anchor, num_classes+1]
        # y_pred [batch_size, num_anchor, num_classes]
        labels         = y_true[:, :, :-1]
        anchor_state   = y_true[:, :, -1]  # -1 是需要忽略的, 0 是背景, 1 是存在目标
        classification = y_pred
        # 找出存在目标的先验框
        indices_for_object        = backend.where(keras.backend.equal(anchor_state, 1))
        labels_for_object         = backend.gather_nd(labels, indices_for_object)
        classification_for_object = backend.gather_nd(classification, indices_for_object)
        # 计算每一个先验框应该有的权重
        alpha_factor_for_object = keras.backend.ones_like(labels_for_object) * alpha
        alpha_factor_for_object = backend.where(keras.backend.equal(labels_for_object, 1), alpha_factor_for_object, 1 - alpha_factor_for_object)
        focal_weight_for_object = backend.where(keras.backend.equal(labels_for_object, 1), 1 - classification_for_object, classification_for_object)
        focal_weight_for_object = alpha_factor_for_object * focal_weight_for_object ** gamma
        # 将权重乘上所求得的交叉熵
        cls_loss_for_object = focal_weight_for_object * keras.backend.binary_crossentropy(labels_for_object, classification_for_object)
        # 找出实际上为背景的先验框
        indices_for_back        = backend.where(keras.backend.equal(anchor_state, 0))
        labels_for_back         = backend.gather_nd(labels, indices_for_back)
        classification_for_back = backend.gather_nd(classification, indices_for_back)
        # 计算每一个先验框应该有的权重
        alpha_factor_for_back = keras.backend.ones_like(labels_for_back) * (1 - alpha)
        focal_weight_for_back = classification_for_back
        focal_weight_for_back = alpha_factor_for_back * focal_weight_for_back ** gamma
        # 将权重乘上所求得的交叉熵
        cls_loss_for_back = focal_weight_for_back * keras.backend.binary_crossentropy(labels_for_back, classification_for_back)
        # 标准化，实际上是正样本的数量
        normalizer = tf.where(keras.backend.equal(anchor_state, 1))
        normalizer = keras.backend.cast(keras.backend.shape(normalizer)[0], keras.backend.floatx())
        normalizer = keras.backend.maximum(keras.backend.cast_to_floatx(1.0), normalizer)
        # 将所获得的loss除上正样本的数量
        cls_loss_for_object = keras.backend.sum(cls_loss_for_object)
        cls_loss_for_back = keras.backend.sum(cls_loss_for_back)
        # 总的loss
        loss = (cls_loss_for_object + cls_loss_for_back)/normalizer
        return loss
    return _focal