Skip to content

Image segmentation on a subset of CamVid dataset using VGG-16 Feature Extractor followed by FCN-8 decoder

Notifications You must be signed in to change notification settings

LucienCastle/FCN8-VGG16Backbone

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

15 Commits
 
 
 
 

Repository files navigation

FCN8-VGG16Backbone

Image segmentation on a subset of CamVid dataset using VGG-16 Feature Extractor followed by FCN-8 decoder

Dataset

The dataset images and annotations. The images contain the video frames while the annotations contain the pixel-wise label maps. Each label map has the shape (height, width , 1) with each point in this space denoting the corresponding pixel's class. Classes are in the range [0, 11] (i.e. 12 classes) and the pixel labels correspond to these classes:

Value Class Name
0 sky
1 building
2 column/pole
3 road
4 side walk
5 vegetation
6 traffic light
7 fence
8 vehicle
9 pedestrian
10 byciclist
11 void

Model

VGG-16 Network

The backbone used is VGG-16, without the fully connected layers, initialized with pretrained weights. The architecture of a VGG-16 network is as follows:
vgg-16

FCN-8

FCN-8 is a fully convolutional network which upsamples the feature extracted by encoder and creates a pixel-wise labelmap fcn-8

Metrics

Metrics used are IoU and Dice Score.


A smoothing factor can be added in both numerator and denominator to avoid 0 division

About

Image segmentation on a subset of CamVid dataset using VGG-16 Feature Extractor followed by FCN-8 decoder

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published