Add FPN connections based on the model described in the FPN paper. fpn_output_blobs is in reversed order: e.g [fpn5, fpn4, fpn3, fpn2] simila

A class representing a COCO json dataset.

A simple timer.

Track a series of values and provide access to smoothed values over a window or the global series average.

Track vital training statistics.

v1upXconvs design: X * (conv 3x3), convT 2x2.

r"""Wraps another sampler to yield a mini-batch of indices. Args: sampler (Sampler): Base sampler. batch_size (int): Size of mini-

Merge RPN proposals generated at multiple FPN levels and then distribute those proposals to their appropriate FPN levels. An anchor at one FPN

v1upXconvs design: X * (conv 3x3), convT 2x2, with GroupNorm

v1upXconvs design: X * (conv 3x3), convT 2x2, with GroupNorm

v1upXconvs design: X * (conv 3x3), convT 2x2, with GroupNorm

Add RPN outputs to a single scale model (i.e., no FPN).

Add a top-down lateral module.

A simple channel-wise affine transformation operation

Bilinear interpolation in space of scale. Takes input of NxKxHxW and outputs NxKx(sH)x(sW), where s:= up_scale Adapted from the CVPR'15 FCN

r"""Implements data parallelism at the module level. This container parallelizes the application of the given module by splitting the input a

Bottleneck Residual Block

Add RPN on FPN specific outputs.

Mask R-CNN keypoint specific outputs: keypoint heatmaps.

v0up design: conv5, deconv 2x2 (no weight sharing with the box head).

Use a ResNet "conv5" / "stage5" head for mask prediction. Weights and computation are shared with the conv5 box head. Computation can only be

Mask R-CNN specific outputs: either mask logits or probs.

Add a ReLU MLP with two hidden layers.

Add a ReLU MLP with two hidden layers.

Add a ReLU MLP with two hidden layers.

Add a X conv + 1fc head, with GroupNorm

Add a X conv + 1fc head, with GroupNorm

Add a X conv + 1fc head, as a reference if not using GroupNorm

Mask R-CNN keypoint head. v1convX design: X * (conv).