tests/utils/losses.py

# coding=utf-8
# Copyright 2022 The IDEA Authors. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

import torch.nn.functional as F


def sigmoid_focal_loss(inputs, targets, num_boxes, alpha: float = 0.25, gamma: float = 2):
    """
    Loss used in RetinaNet for dense detection: https://arxiv.org/abs/1708.02002.

    Args:
        inputs (torch.Tensor): A float tensor of arbitrary shape.
            The predictions for each example.
        targets (torch.Tensor): A float tensor with the same shape as inputs. Stores the binary
            classification label for each element in inputs
            (0 for the negative class and 1 for the positive class).
        num_boxes (int): The number of boxes.
        alpha (float, optional): Weighting factor in range (0, 1) to balance
            positive vs negative examples. Default: 0.25.
        gamma (float): Exponent of the modulating factor (1 - p_t) to
            balance easy vs hard examples. Default: 2.

    Returns:
        torch.Tensor: The computed sigmoid focal loss.
    """
    inputs = inputs.float()
    targets = targets.float()
    prob = inputs.sigmoid()
    ce_loss = F.binary_cross_entropy_with_logits(inputs, targets, reduction="none")
    p_t = prob * targets + (1 - prob) * (1 - targets)
    loss = ce_loss * ((1 - p_t) ** gamma)

    if alpha >= 0:
        alpha_t = alpha * targets + (1 - alpha) * (1 - targets)
        loss = alpha_t * loss

    return loss.mean(1).sum() / num_boxes


def dice_loss(inputs, targets, num_boxes):
    """
    Compute the DICE loss, similar to generalized IOU for masks
    Args:
        inputs (torch.Tensor): A float tensor of arbitrary shape.
            The predictions for each example.
        targets (torch.Tensor):
            A float tensor with the same shape as inputs. Stores the binary
            classification label for each element in inputs
            (0 for the negative class and 1 for the positive class).
        num_boxes (int): The number of boxes.
    Return:
        torch.Tensor: The computed dice loss.
    """
    inputs = inputs.sigmoid()
    inputs = inputs.flatten(1)
    numerator = 2 * (inputs * targets).sum(1)
    denominator = inputs.sum(-1) + targets.sum(-1)
    loss = 1 - (numerator + 1) / (denominator + 1)
    return loss.sum() / num_boxes