semantic-segmentation-adas-0001

Use Case and High-Level Description

This is a segmentation network to classify each pixel into 20 classes:

  • road
  • sidewalk
  • building
  • wall
  • fence
  • pole
  • traffic light
  • traffic sign
  • vegetation
  • terrain
  • sky
  • person
  • rider
  • car
  • truck
  • bus
  • train
  • motorcycle
  • bicycle
  • ego-vehicle

Example

Specification

Metric Value
Image size 2048x1024
GFlops 58.572
MParams 6.686
Source framework Caffe*

Accuracy

The quality metrics calculated on 2000 images:

Label IOU
mean 0.6907
Road 0.910379
Sidewalk 0.630676
Building 0.860139
Wall 0.424166
Fence 0.592632
Pole 0.559078
Traffic Light 0.654779
Traffic Sign 0.648217
Vegetation 0.882593
Terrain 0.620521
Sky 0.976889
Person 0.711653
Rider 0.612787
Car 0.877892
Truck 0.674829
Bus 0.743752
Train 0.358641
Motorcycle 0.600701
Bicycle 0.622246
Ego-Vehicle 0.852932
  • IOU=TP/(TP+FN+FP), where:
    • TP - number of true positive pixels for given class
    • FN - number of false negative pixels for given class
    • FP - number of false positive pixels for given class

Inputs

The blob with BGR image and the shape 1, 3, 1024, 2048 in the format B, C, H, W, where:

  • B – batch size
  • C – number of channels
  • H – image height
  • W – image width

Outputs

The net output is a blob with the shape 1, 1, 1024, 2048 in the format B, C, H, W. It can be treated as a one-channel feature map, where each pixel is a label of one of the classes.

Legal Information

[*] Other names and brands may be claimed as the property of others.