Exercise
Create a deep learning (segmentation) model which fulfills the below:

1. Road, sidewalk, cars, trees segmentation which inferences on video and real time webcam feed. (you can use public available datasets for the same)
2. The created model should be converted to Tflite format and must run on a mobile platform.
3. FPS required at least 7-8.
4. You can use TensorFlow or Pytorch to train the model and convert it to mobile compatible format.
5. Accuracy needed at least 75-80% for segmentation.