Social Data UK meeting up at the Shard (Gallup 18th Floor)
25/05/2016
Presenter from Brandwatch and Pulsar
Not a bad view – first time up the #shard for “analysing images in social data” @Meetup #SocialDataUK pic.twitter.com/4siTCkNcjT
— Alice Data (@alice_data) May 25, 2016
Why is Visual Listening Important ?
Why do visual listening? brand protection, brand monitoring and consumer insights #SocialDataUK pic.twitter.com/zSTqFRDQui
— Alice Data (@alice_data) May 25, 2016
Why bother?
Every day millions of images are shares but @Brandwatch is working on logo pattern recognition #SocialDataUK pic.twitter.com/M5TmijHIgp
— Alice Data (@alice_data) May 25, 2016
But there are constraints – misidentified logo , low frequency and randomness #socialdatauk pic.twitter.com/rbBCSCqmsX
— Alice Data (@alice_data) May 25, 2016
Monitoring the good & bad of brand being shared through visual listening #SocialDataUk pic.twitter.com/ueycgF1l8y
— Alice Data (@alice_data) May 25, 2016
Deep learning can be used to help remove noise & rude shares from public social walls #SocialDataUK
— Alice Data (@alice_data) May 25, 2016
Next Step on pattern recognition to machine learning
History lesson-it’s taken us 1000yrs for us get back to pics of day-to-day life (painting to photos) #socialdatauk pic.twitter.com/7aIERYLf4Q
— Alice Data (@alice_data) May 25, 2016
“We share 1.8billion images on social media everyday” via @abc3d @Pulsar_Social #socialdatauk pic.twitter.com/YWzSEqdCFD
— Alice Data (@alice_data) May 25, 2016
Key to Social images analysis : explore, mine, visualise, contextualise #socialdatauk via @abc3d pic.twitter.com/A1xzA0oWRs
— Alice Data (@alice_data) May 25, 2016
Key to Social images analysis : explore, mine, visualise, contextualise #socialdatauk via @abc3d pic.twitter.com/A1xzA0oWRs
— Alice Data (@alice_data) May 25, 2016
Types of Image analysis
Types of image analysis via @abc3d @Pulsar_Social #socialdatauk pic.twitter.com/MEpcCLEC5g
— Alice Data (@alice_data) May 25, 2016
Nothing is perfect
Image analysis-there are problems…too much for the machine to understand, lack of interface&storage #SocialDataUK pic.twitter.com/gx39W7Lsc0
— Alice Data (@alice_data) May 25, 2016
Application ideas
– Recognise brands and sponsorship
– Recognise Building or objects
– Automated Object identification and content (good training needed)