Diet-Aware Dining Table – Observing Dietary Behaviors over Tabletop Surface Shih-yen Liu, Cheryl Chen, Tung-yun Lin, Polly Huang National Taiwan University Keng-hao Chang Hao-(hua) Chu Jane Yung-jen Hsu
Diet-aware Dining Table, NTU A story - motivation Video [Script]: –A man wants to control weight –Doctor asks him to report his dietary habits –Questionnaire is cumbersome, awkward –Then he uses our table, everything is so easy…
Diet-aware Dining Table, NTU Pervasive Healthcare - We are what we eat It’s hard Shopping receipt scanner, Mankoff et al., Ubicomp 2002 –Analyze the purchased food items of a whole family –It cannot track individual intake Analysis of Chewing Sounds for Dietary Monitoring, Amft et al., Ubicomp 2005 –Infer food intake by chewing sound –Ambiguity
Diet-aware Dining Table, NTU But, we try differently Smart object approach –Instrument everyday dining tables –Not blind to what happened above the surface Features: –Natural interaction –Multi-users but in individual level Observed Interactions?
Diet-aware Dining Table, NTU Target interactions Consume food from the “personal” containers Where the food comes from? Transferred from the share containers to personal containers
Diet-aware Dining Table, NTU Demonstration
Diet-aware Dining Table, NTU The table design – what’s the magic? Two sensor surfaces –RFID & Weight RFID – what – RFID-tagged containers Weight - how much – Weight “change” of dietary behaviors Cell division –Concurrent person-container interactions RFID Antenna Load Sensor
Diet-aware Dining Table, NTU Weight consistency principle Transfer tea Drink tea Weight Decrease ofWeight Increase ofWeight Decrease ofWeight Increase of
Diet-aware Dining Table, NTU 1. Transfer Tea Bob pours tea from the tea pot to personal cup Pour tea? Weight increases w 2. Pick up tea pot. RFID tag disappears Weight decreases w 1 Put on tea pot. RFID tag appears Weight increases w 1 - w 2 w1w1 w2w2 w 1 - w 2 w2w2 Pour tea by match !
Diet-aware Dining Table, NTU w 1 -w 2 2. Drink Tea Bob drinks tea Pick up cup. RFID tag disappears. Weight decreases w 1. Put on cup. Drink tea RFID tag appears. Weight increases w 2. w1w1 w 2 Drink tea by identify “Bob”
Diet-aware Dining Table, NTU 3. Complex Example Bob pours tea & Alan cuts cake Pour tea? Cut cake? Weight change w Pour tea Weight increases w 1 Cut cake Weight decreases w 2
Diet-aware Dining Table, NTU Method summary Transfer interactions –Match weight Eat interactions –Identify personal container Concurrent interactions –Divide cells
Diet-aware Dining Table, NTU Experiments Chinese-style dinner scenario with three users No hands, utensils on the table 30 min, 100 transfer events, 60 eat events Behavior Recognition Accuracy: 83.33% –Transfer: 81.99% –Eat: 88.33% Weight Accuracy: % A B C
Diet-aware Dining Table, NTU Experiment Discussion Causes of misses Touching tableEat without TransferWeight Ambiguity 10 g
Diet-aware Dining Table, NTU Conclusion Diet-aware dining table –A smart object and a smart surface –Support natural user interaction –fine-grained dietary tracking at individual level A nice first step in such direction. –80% accuracy. The whole problem can be explored more deeply.
Diet-aware Dining Table, NTU Future work To improve recognition accuracy To relax constraints Just-in-time persuasive technology –To encourage balanced diet
Questions & Answers Thank you! Keng-hao Chang National Taiwan University