comscore Twitter now uses machine learning to auto-crop images smartly
News

Twitter now uses machine learning to auto-crop images smartly

Twitter is using advanced neural networks to crop images to their most interesting part

  • Published: January 28, 2018 12:48 PM IST
Twitter ML feat

Twitter is now using machine learning to automatically crop picture previews to their most interesting part. The use of machine learning and neural networks for image recognition as become common with both Google and Facebook heavily deploying the technology on their platforms. Now, Twitter is also joining with its own implementation with a single aim of improving user experience. Also Read - Happy Dusshera 2021 messages, images, stickers, quotes: How to create, send Happy Vijayadashami greetings via WhatsApp

Also Read - Twitter may soon pay Spaces creators under its new accelerator programme

Twitter explains that it has been working on this technology for a while now and describes the methods in a detailed blog post. Machine Learning Researcher Lucas Theis and Software Engineer Zehan Wang explain that the micro-blogging platform previously used face detection to focus the image on the most prominent face it could find. However, since all the images don’t have faces, the company found the approach didn’t work with all kinds of pictures especially objects and cats. Also Read - Your Twitter experience is going to change, ads tested for replies

Twitter has now tweaked its algorithm with what it calls as cropping using saliency. The platform now crops images based on whatever is interesting and not just limited to faces. Twitter defines the most interesting part as a region having high saliency where a person is likely to look at when freely viewing the image.

“In general, people tend to pay more attention to faces, text, animals, but also other objects and regions of high contrast.”

Twitter now uses this data to train its neural networks and other algorithms to predict what people might want to look at rather than limiting it to prominent face. Engineers at Twitter note that recent advances in machine learning has made saliency prediction a lot better. The major challenge for the company being the sllow processing speed of neural networks deployed to predict saliency, making them not suitable to run in production.

However, Twitter optimized the neural network s implementation by using two techniques that help reduce the size and hence the computational requirements. Twitter explains using a technique called knowledge distillation to train a similar network in order to imitate the slower but more powerful network. The second is a pruning technique to remove feature maps of the neural network which were costly to compute but contributed in little to performance.

In raw performance terms, Twitter claims to have made its neural networks 10x faster than a vanilla implementation. The update is in the process of being rolled out to everyone accessing the service on web, iOS and Android apps. So, next time if you see a cat prominently on your Twitter feed then it is powered by these very neural networks.

For the latest tech news across the world, latest PC and Mobile games, tips & tricks, top-notch gadget reviews of most exciting releases follow BGR India’s Facebook, Twitter, subscribe our YouTube Channel. Also follow us on  Facebook Messenger for latest updates.
  • Published Date: January 28, 2018 12:48 PM IST



new arrivals in india

Realme 8i
Realme 8i

13,999

Realme 8s 5G
Realme 8s 5G

17,999

Xiaomi Redmi 10 Prime
Xiaomi Redmi 10 Prime

12,499

Samsung Galaxy A52s 5G
Samsung Galaxy A52s 5G

35,999

Samsung Galaxy M32 5G
Samsung Galaxy M32 5G

20,999

Vivo Y33s
Vivo Y33s

17,990

Realme C21Y
Realme C21Y

8,999

Vivo Y21
Vivo Y21

15,490

Realme GT Master Edition
Realme GT Master Edition

25,999

Realme GT 5G
Realme GT 5G

37,999

Itel A48
Itel A48

6,399

Motorola Edge 20
Motorola Edge 20

29,999

Motorola Edge 20 Fusion
Motorola Edge 20 Fusion

21,499

Samsung Galaxy Z Flip 3
Samsung Galaxy Z Flip 3

84,999

Samsung Galaxy Z Fold 3
Samsung Galaxy Z Fold 3

1,49,999

Samsung Galaxy A12 (Exynos)
Samsung Galaxy A12 (Exynos)

13,999

Nokia C20 Plus
Nokia C20 Plus

8,999

Vivo Y53s
Vivo Y53s

19,490

Tecno Pova 2
Tecno Pova 2

10,999

Infinix Smart 5A
Infinix Smart 5A

6,499

Micromax In 2b
Micromax In 2b

8,999

Vivo Y72 5G
Vivo Y72 5G

20,990

Tecno Camon 17
Tecno Camon 17

12,999

Tecno Camon 17 Pro
Tecno Camon 17 Pro

16,999

Realme C11 2021
Realme C11 2021

6,999

Oppo Reno6 Pro 5G
Oppo Reno6 Pro 5G

39,990

Oppo Reno6 5G
Oppo Reno6 5G

29,990

Samsung Galaxy M21 2021
Samsung Galaxy M21 2021

12,499

OnePlus Nord 2
OnePlus Nord 2

27,999

Poco F3 GT
Poco F3 GT

27,999

Samsung Galaxy A22 5G
Samsung Galaxy A22 5G

19,999

Xiaomi Redmi Note 10T 5G
Xiaomi Redmi Note 10T 5G

13,999

Samsung Galaxy F22
Samsung Galaxy F22

12,499

Xiaomi Mi 11 Lite
Xiaomi Mi 11 Lite

21,999

Infinix Note 10 Pro
Infinix Note 10 Pro

16,999

Infinix Note 10
Infinix Note 10

10,999

Vivo Y73
Vivo Y73

20,990

OnePlus Nord CE 5G
OnePlus Nord CE 5G

22,999

iQOO Z3
iQOO Z3

19,990

Realme C25s
Realme C25s

9,999

Poco M3 Pro 5G
Poco M3 Pro 5G

13,999

Realme X7 Max 5G
Realme X7 Max 5G

26,999

Oppo F19
Oppo F19

18,990

Motorola Moto G40 Fusion
Motorola Moto G40 Fusion

13,999

POCO M2 Reloaded
POCO M2 Reloaded

9,499

OPPO A74 5G
OPPO A74 5G

17,990

Oppo A53s 5G
Oppo A53s 5G

14,990

Vivo V21 5G
Vivo V21 5G

29,990

Realme C25
Realme C25

9,499

Realme C21
Realme C21

7,999

Realme C20
Realme C20

6,799

Motorola Moto G60
Motorola Moto G60

17,999

iQOO 7
iQOO 7

31,990

Samsung Galaxy M42 5G
Samsung Galaxy M42 5G

21,999

Xiaomi Mi 11 Ultra
Xiaomi Mi 11 Ultra

69,999

Xiaomi Mi 11X Pro 5G
Xiaomi Mi 11X Pro 5G

39,999

Xiaomi Mi 11X
Xiaomi Mi 11X

29,999

Realme 8 5G
Realme 8 5G

13,999

Samsung Galaxy F02s
Samsung Galaxy F02s

8,999

Samsung Galaxy F12
Samsung Galaxy F12

10,999

POCO X3 Pro
POCO X3 Pro

18,999

Realme 8 Pro
Realme 8 Pro

17,999

Realme 8
Realme 8

14,999

Vivo X60 Pro Plus
Vivo X60 Pro Plus

69,990

Vivo X60 Pro
Vivo X60 Pro

49,990

Vivo X60
Vivo X60

37,990

OnePlus 9 Pro 5G
OnePlus 9 Pro 5G

64,999

OnePlus 9R 5G
OnePlus 9R 5G

39,999

OnePlus 9 5G
OnePlus 9 5G

49,999

Samsung Galaxy A72
Samsung Galaxy A72

34,999

Samsung Galaxy A52
Samsung Galaxy A52

26,499

Micromax In 1
Micromax In 1

10,499

Asus ROG Phone 5
Asus ROG Phone 5

49,999

Samsung Galaxy M12
Samsung Galaxy M12

10,999

Motorola Moto G30
Motorola Moto G30

10,999

Motorola Moto G10 Power
Motorola Moto G10 Power

9,999

Oppo F19 Pro Plus 5G
Oppo F19 Pro Plus 5G

25,990

Oppo F19 Pro
Oppo F19 Pro

21,490

Xiaomi Redmi Note 10 Pro Max
Xiaomi Redmi Note 10 Pro Max

18,999

Xiaomi Redmi Note 10 Pro
Xiaomi Redmi Note 10 Pro

15,999

Xiaomi Redmi Note 10
Xiaomi Redmi Note 10

11,999

Realme Narzo 30A
Realme Narzo 30A

8,999

Realme Narzo 30 Pro
Realme Narzo 30 Pro

16,999

Infinix Smart 5
Infinix Smart 5

7,199

Samsung Galaxy F62
Samsung Galaxy F62

23,999

Samsung Galaxy A12
Samsung Galaxy A12

12,999

Nokia 5.4
Nokia 5.4

13,999

Nokia 3.4
Nokia 3.4

11,999

Realme X7 Pro 5G
Realme X7 Pro 5G

29,999

Realme X7
Realme X7

19,999

Vivo Y31
Vivo Y31

16,490

Oppo Reno5 Pro 5G
Oppo Reno5 Pro 5G

35,990

Samsung Galaxy S21 Ultra 5G
Samsung Galaxy S21 Ultra 5G

1,05,999

Samsung Galaxy S21 Plus 5G
Samsung Galaxy S21 Plus 5G

81,999

Samsung Galaxy S21 5G
Samsung Galaxy S21 5G

69,999

Vivo Y12s
Vivo Y12s

9,990

Vivo Y51A
Vivo Y51A

17,990

Samsung Galaxy M02s
Samsung Galaxy M02s

8,999

Xiaomi Mi 10i
Xiaomi Mi 10i

21,999

Oppo A15s
Oppo A15s

11,490

Best Sellers