Reshaping happens during the call to forward

Reshaping happens during the call to forward.

Reshaping happens during the call to forward

Reshaping happens during the call to forward.

Clip boxes inside image boundaries

Reshaping happens during the call to forward.

Reshaping happens during the call to forward.

Dispatch to either CPU or GPU NMS implementations.

Get an imdb (image database) by name.

Convert a list of images into a network input. Assumes images are already prepared (means subtracted, BGR order, ...).

invert bounding box transform apply delta on anchors to get transformed proposals

Generate anchor (reference) windows by enumerating aspect ratios X scales wrt a reference (0, 0, 15, 15) window.

Mean subtract and scale an image for use in a blob.

Unmap a subset of item (data) back to the original set of items (of size count)

Randomly permute the training roidb.

Compute bounding-box regression targets for an image Parameters: ----------- ex_rois: ROIs from external source (anchors or proposals

Compute bbox regression targets of external rois with respect to gt rois

Bounding-box regression targets (bbox_target_data) are stored in a compact form N x (class, tx, ty, tw, th) This function expands those targe

This function calculate the intersection part of a external box and gt_box, mask it according to gt_mask Args: ex_box: external

This function calculate region IOU when masks are inside different boxes Returns: intersection over unions of this two masks

Given a vector of widths (ws) and heights (hs) around a center (x_ctr, y_ctr), output a set of anchors (windows).

Randomly permute the training roidb.

Return width, height, x center, and y center for an anchor (window).

Load a config file and merge it into the default options.

Compute bounding-box regression targets for an image.

Remove all boxes with any side smaller than min_size.

Return the directory where experimental artifacts are placed. A canonical path is built using the name from an imdb and a network (if

A wrapper function, note we already know the class of boxes and masks Args: masks: ~ 300 x 21 x 21 boxes: ~ 300 x 4 s

Return the database of ground-truth regions of interest. This function loads/saves from/to a cache file to speed up future calls.

Converts an image into a network input. Arguments: im (ndarray): a color image in BGR order Returns: blob (ndarray): a data bl

Take a snapshot of the network after unnormalizing the learned bounding-box regression weights. This enables easy use at test-time.

Compute VOC AP given precision and recall. If use_07_metric is true, uses the VOC 07 11 point method (default:False). Args: rec: