get_boxes: Transforms 'Yolo3' predictions into valid boxes.

Description Usage Arguments Value

View source: R/yolo3_predict.R

Description

Transforms 'Yolo3' predictions into valid boxes.

Usage

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
get_boxes(
  preds,
  anchors,
  labels,
  obj_threshold = 0.6,
  nms = TRUE,
  nms_threshold = 0.6,
  correct_hw = FALSE,
  image_h = NULL,
  image_w = NULL
)

Arguments

preds

yolo3 model predictions.

anchors

Prediction anchors. For exact format check coco_anchors.

labels

Character vector containing class labels. For example coco_labels.

obj_threshold

Minimum objectness score. Must be in range '[0, 1]'. All boxes with objectness score less than 'obj_threshold' will be filtered out.

nms

Logical. Should 'Non-Maximum-Suppression' be applied.

nms_threshold

'Non-Maximum-Suppression' threshold.

correct_hw

Logical. Should height/width rescaling of bounding boxes be applied. If 'TRUE' 'xmin/xmax' coordinates are multiplied by 'image_w' and 'ymin/ymax' coordinates are multiplied by 'image_h'.

image_h

Rescaling factor for 'ymin/ymax' box coordinates.

image_w

Rescaling factor for 'xmin/xmax' box coordinates.

Value

List of 'data.frames' containing bounding box coordinates and objectness/class scores.


maju116/platypus documentation built on Oct. 18, 2020, 9:40 a.m.