Feature pyramid network

As explained in Deep Learning with R, Third Edition

A feature pyramid network is a neural-network architecture that combines semantically rich low-resolution features with spatially detailed high-resolution features at multiple scales.

A feature pyramid network (FPN) helps object detectors recognize both small and large objects by combining detailed high-resolution features with semantically rich low-resolution features. section 12.3

An FPN upsamples final convolutional features through progressive Conv2DTranspose layers, then adds lateral connections from same-sized earlier feature maps. section 12.3

Use an FPN through pretrained RetinaNet

A pretrained RetinaNet model exposes FPN-based detection through KerasHub: `detector <- keras_hub$models$ObjectDetector$from_preset("retinanet_resnet50_fpn_v2_coco", bounding_box_format = "rel_xywh")` `predictions <- predict(detector, list(image))` The returned predictions include bounding boxes, confidence scores, labels, and the number of detections. section 12.3

Figure 12.8 An FPN creates semantically interesting feature maps at different scales.

Use the architecture sketch to see how semantically meaningful feature maps are produced at multiple spatial scales. section 12.3

How Deep Learning with R, Third Edition covers it

How deep it goes

Explains feature fusion through upsampling and lateral additions, without deriving training objectives or implementation internals.

Examples

Uses pretrained RetinaNet inference on a Pointillist painting, with bounding-box visualization and COCO labels.

What you can do after

Operate/configure a pretrained RetinaNet detector through KerasHub and inspect its multiscale detection outputs.

How it connects

Connects convolutional neural networks, object detection, RetinaNet, and YOLO through multiscale feature maps.

liveBookmanning.com

Read Deep Learning with R, Third Edition on liveBook Get the book at manning.com

Also explained in

• 

  liveBookmanning.com
  Deep Learning with Python, Third EditionAn architecture that merges multi-scale feature maps from different depths to support detection of objects at multiple sizes.

Compare how 2 books explain feature pyramid network →

Related concepts

• RetinaNetused by · RetinaNet is a one-stage object-detection architecture that uses convolutional feature pyramids with separate classification and bounding-box regression heads.
• feature mapuses · A feature map is a spatially organized array of neural-network activations representing the presence or strength of learned features at different locations.
• Conv2DTransposeuses · Conv2DTranspose is a neural-network layer that applies a learned two-dimensional transposed convolution to increase the spatial resolution of feature maps.
• upsamplingimplements · Upsampling is a process that increases the number, density, or spatial or temporal resolution of samples or values in a dataset or representation by adding, duplicating, or estimating them.

This explainer was generated by Manning's knowledge graph from the text of Deep Learning with R, Third Edition. The section links point to the parts of the book it draws on.