Huawei's Real-Time Video Multi-instance Segmentation

youtube.com/watch?v=Tipyl6Gzqrs

How is this even possible on a mobile chip?

Attached: huawei-multi-instances-segmentation-kirin-1.jpg (1022x574, 73K)

>Huawei
Tech stolen from the yankees

using a time machine?

shame they cant tie the audio track to the individual musician so we can give a bit more oomph to that sweet sweet sax

that'll be the mate 40. They ain't going to empty their nuts in one go

Holy fucking shit, there's no way it works that well on a fucking phone. Wait for demos. Either that or Huawei singlehandedly BTFO'd the entire computing industry in 1 go.

What do you mean? Mobile chips are extremely powerful. The fact you're used to hacked together web code is why you think it should be laggy

Yankee coder here

We don't have anything like this

What exactly is impressive about overlaying video? Commodore Amiga did it in 1987

It's the detecting edges of objects that's impressive

It's not going to work as well as they're advertising it, I guarantee it

Huawei needs better PR. Imagine if Apple releases something like that. it will be labeled as revolutionary instantly.

Yes I think most people here realize that

fuck off chinkshill

Attached: Screenshot_20190907-192456~01.png (757x107, 19K)

Imagine an ad with only haired blue eyed white people in it. Asians have balls.

Well, Europe is their largest market second only to China. Almost everyone I know (in Poland) has a chinese phone. Those poorer have Xiaomis and those more affluent carry Huaweis.

Because the original footage is with a green screen, but that's in the fine print.

Imagine the military implications.

Isn't this the same company that has been REPEATEDLY caught using DSLR photos to advertise their phone camera? Why the fuck would I give two shits about a Huawei ad?

Attached: 1561314341471.gif (300x275, 720K)

wow... some chink copied pasta'd a github deeep learning algo, impressive

>chroma keying and image layering/resizing
>DEEP LEARNING
gtfo zoomer