Displayed are triplets annotated with human referential utterances. A neural listener, trained with chairs, is used to color-code attention-wise the utterances and score each object. Underlined tokens where out-of-the-vocabulary and were ignored by the listener.
| Distractor-A | Distractor-B | Target |
|
|
|
|
Human Speaker: the target is a tall chair Correctly guessed: True Confidences: 0.00, 0.00, 1.00 |
||
| Distractor-A | Distractor-B | Target |
|
|
|
|
|
Human Speaker:
this is seating for one person , with narrow Correctly guessed: True Confidences: 0.00, 0.01, 0.99 |
||
| Distractor-A | Distractor-B | Target |
|
|
|
|
Human Speaker: l - shaped with pillows Correctly guessed: True Confidences: 0.01, 0.00, 0.99 |
||
| Distractor-A | Distractor-B | Target |
|
|
|
|
|
Human Speaker:
it is in an l shape there is an arm rest on Correctly guessed: True Confidences: 0.00, 0.00, 1.00 |
||
| Distractor-A | Distractor-B | Target |
|
|
|
|
Human Speaker:
it looks kind of like a park bench with wooden Correctly guessed: True Confidences: 0.00, 0.01, 0.99 |
||
| Distractor-A | Distractor-B | Target |
|
|
|
|
|
Human Speaker: the target has railings in the back Correctly guessed: True Confidences: 0.00, 0.00, 1.00 |
||
| Distractor-A | Distractor-B | Target |
|
|
|
|
Human Speaker: this sofa has rounded back and corners Correctly guessed: True Confidences: 0.00, 0.01, 0.99 |
||
| Distractor-A | Distractor-B | Target |
|
|
|
|
|
Human Speaker:
it has a fairly low back which goes continuous Correctly guessed: False Confidences: 0.14, 0.53, 0.33 |
||
| Distractor-A | Distractor-B | Target |
|
|
|
|
Human Speaker: the target has two back pillows Correctly guessed: True Confidences: 0.01, 0.00, 0.99 |
||
| Distractor-A | Distractor-B | Target |
|
|
|
|
|
Human Speaker:
this couch has two separate back cushions Correctly guessed: True Confidences: 0.12, 0.02, 0.86 |
||