In this project, you'll label the pixels of a road in images using a Fully Convolutional Network (FCN).
The design implementation here is very close to this paper on Fully Convolutional Networks for Semantic Segmentation. .
In this we are using a pre-trained VGG-16 network and replaced the final layer fully connected layer to a 1x1 CNN. The depth setting was equal to the number of classes we have and in this case we are finding pixels which are roads with non roads .
Then performance of the network was inproved with the addition of skip connection. Here we do a element wise add with the vgg 3rd and 4th layer to the 1x1 convolved layer 7 and layer 4 respectively.
- Loss function: Cross entropy
- Optimizer: Adam optimizer
- learning rate: 0.0001
- key-prob:0.5
- epoach :50
- batch :10
###Ouput
Here is the output of the FCN to find the roads. As you can see the green highlighted area are the pixels that are supposed to be road surface to the FCN.
Make sure you have the following is installed:
Download the Kitti Road dataset from here. Extract the dataset in the data
folder. This will create the folder data_road
with all the training a test images.
Implement the code in the main.py
module indicated by the "TODO" comments.
The comments indicated with "OPTIONAL" tag are not required to complete.
Run the following command to run the project:
python main.py
Note If running this in Jupyter Notebook system messages, such as those regarding test status, may appear in the terminal rather than the notebook.
- Ensure you've passed all the unit tests.
- Ensure you pass all points on the rubric.
- Submit the following in a zip file.
helper.py
main.py
project_tests.py
- Newest inference images from
runs
folder
A well written README file can enhance your project and portfolio. Develop your abilities to create professional README files by completing this free course.