أرسل هذا في رسالة قصيرة: Effective representations for road scene understanding with scalable learning