Skip to content

Latest commit

 

History

History

0-old

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
assets
assets
 
 
nets
nets
 
 
README.md
README.md
 
 
v6_01_preprocess.py
v6_01_preprocess.py
 
 
v6_02_tfrecord.py
v6_02_tfrecord.py
 
 
v6_03_cnn_train.py
v6_03_cnn_train.py
 
 
v6_04_cnn_test.py
v6_04_cnn_test.py
 
 
v6_05_resnet_train.py
v6_05_resnet_train.py
 
 
v6_06_resnet_test.py
v6_06_resnet_test.py
 
 
v6_07_accuracy.py
v6_07_accuracy.py
 
 
v6_08_mosaic.py
v6_08_mosaic.py
 
 

README.md

Compare simple CNN and ResNet for building extraction with tensorflow. Out of date, but interesting trial.

Data

Massachusetts building dataset

ISPRS dataset

Each big tile (e.g. 15001500) is overlap-cropped to 6464 patches with stride 16. The center 16*16 sub-patch is the label of the patch.

Model

At this time, no fully convolutional networks were used. The model (CNN or ResNet) has a final fully connected layer with output shape (B,256,2). In other words, the center sub-patch is flattened and each pixel is classified.

Result

Interesting sawtooth effect. From left to right: RGB image, groud truth, prediction.

Discussion

note: Not yet fully convolutional network based sementic segmentation, out of date, but interesting trial.