Object Detection

These are common losses used in object detection.

---

GIoU Loss

pl_bolts.losses.object_detection.giou_loss(preds, target)

Calculates the generalized intersection over union loss.

It has been proposed in Generalized Intersection over Union: A Metric and A Loss for Bounding Box Regression.

Parameters

• preds (Tensor) – an Nx4 batch of prediction bounding boxes with representation [x_min, y_min, x_max, y_max]

• target (Tensor) – an Mx4 batch of target bounding boxes with representation [x_min, y_min, x_max, y_max]

Example

>>> import torch
>>> from pl_bolts.losses.object_detection import giou_loss
>>> preds = torch.tensor([[100, 100, 200, 200]])
>>> target = torch.tensor([[150, 150, 250, 250]])
>>> giou_loss(preds, target)
tensor([[1.0794]])

Return type

Tensor

Returns

GIoU loss in an NxM tensor containing the pairwise GIoU loss for every element in preds and target, where N is the number of prediction bounding boxes and M is the number of target bounding boxes

---

IoU Loss

pl_bolts.losses.object_detection.iou_loss(preds, target)

Calculates the intersection over union loss.

Parameters

• preds (Tensor) – batch of prediction bounding boxes with representation [x_min, y_min, x_max, y_max]

• target (Tensor) – batch of target bounding boxes with representation [x_min, y_min, x_max, y_max]

Example

>>> import torch
>>> from pl_bolts.losses.object_detection import iou_loss
>>> preds = torch.tensor([[100, 100, 200, 200]])
>>> target = torch.tensor([[150, 150, 250, 250]])
>>> iou_loss(preds, target)
tensor([[0.8571]])

Return type

Tensor

Returns

IoU loss

---

Reinforcement Learning

These are common losses used in RL.

---

DQN Loss

pl_bolts.losses.rl.dqn_loss(batch, net, target_net, gamma=0.99)

Calculates the mse loss using a mini batch from the replay buffer.

Parameters

• batch (Tuple[Tensor, Tensor]) – current mini batch of replay data

• net (Module) – main training network

• target_net (Module) – target network of the main training network

• gamma (float) – discount factor

Return type

Tensor

Returns

loss

Double DQN Loss

pl_bolts.losses.rl.double_dqn_loss(batch, net, target_net, gamma=0.99)

Calculates the mse loss using a mini batch from the replay buffer. This uses an improvement to the original DQN loss by using the double dqn. This is shown by using the actions of the train network to pick the value from the target network. This code is heavily commented in order to explain the process clearly.

Parameters

• batch (Tuple[Tensor, Tensor]) – current mini batch of replay data

• net (Module) – main training network

• target_net (Module) – target network of the main training network

• gamma (float) – discount factor

Return type

Tensor

Returns

loss

Per DQN Loss

pl_bolts.losses.rl.per_dqn_loss(batch, batch_weights, net, target_net, gamma=0.99)

Calculates the mse loss with the priority weights of the batch from the PER buffer.

Parameters

• batch (Tuple[Tensor, Tensor]) – current mini batch of replay data

• batch_weights (List) – how each of these samples are weighted in terms of priority

• net (Module) – main training network

• target_net (Module) – target network of the main training network

• gamma (float) – discount factor

Return type

Tuple[Tensor, ndarray]

Returns

loss and batch_weights