Multi-task deep convolutional neural networks for efficient and robust traffic lane detection

A method comprising:

• receiving an image of a vehicle environment;
• processing the image using a convolutional neural network including at least one convolutional layer, at least one max-pooling layer, and at least one fully connected layer;
• calculating a prediction regarding whether the image contains a traffic lane marking segment based on outputs generated by the convolutional neural network;
• calculating geometric characteristics of the traffic lane marking segment based on the outputs generated by the convolutional neural network;
• and
• training the convolutional neural network based on the geometric characteristics of the traffic lane marking segment only if the prediction is that the image contains the traffic lane marking segment.

A system comprising:

• one or more cameras disposed on a vehicle;
• one or more processors for controlling operations of the computing device;
• and
• a memory storing data and program instructions used by the one or more processors, wherein the one or more processors are configured to execute instructions stored in the memory to:
• receive an image of a vehicle environment from the one or more cameras;
• process the image using a convolutional neural network including at least one convolutional layer, at least one max-pooling layer, and at least one fully connected layer;
• calculate a prediction regarding whether the image contains a traffic lane marking segment based on outputs generated by the convolutional neural network;
• and
• only if the prediction is that the image contains the traffic lane marking segment, calculate geometric characteristics of the traffic lane marking segment.