(video) Wayve LINGO-1 Show & Tell – Traffic Light

When we ask about the colour of the traffic light, LINGO-1 can accurately describe that “the traffic light is currently red” and simultaneously highlight the traffic light in the image frame, denoted by [A]. Furthermore, LINGO-1 can elaborate on the correct driving response to a red traffic light by stating, “We are waiting for it to turn green before proceeding.” As you can see, the introduction of referential segmentation strengthens the link between language and vision tasks, improving our confidence in and ability to assess the accuracy and relevancy of LINGO-1’s responses.