Commit
•
365b962
1
Parent(s):
ee67b3c
Update README.md
Browse files
README.md
CHANGED
|
@@ -1,7 +1,7 @@
|
|
| 1 |
|
| 2 |
# Visual semantic with BERT-CNN
|
| 3 |
|
| 4 |
-
This model can be used to assign an object-to-caption relatedness score, which is valuable for
|
| 5 |
(1) caption diverse re-ranking, and (2) generate soft labels for caption filtering when scraping text-to-captions from the internet.
|
| 6 |
|
| 7 |
To take advantage of the overlapping between the visual context and the caption, and to extract global information from each visual (i.e., object, scene, etc) we use BERT as an embedding layer followed by a shallow CNN (tri-gram kernel) (Kim, 2014).
|
|
|
|
| 1 |
|
| 2 |
# Visual semantic with BERT-CNN
|
| 3 |
|
| 4 |
+
This model can be used to assign an object-to-caption semantic relatedness score, which is valuable for
|
| 5 |
(1) caption diverse re-ranking, and (2) generate soft labels for caption filtering when scraping text-to-captions from the internet.
|
| 6 |
|
| 7 |
To take advantage of the overlapping between the visual context and the caption, and to extract global information from each visual (i.e., object, scene, etc) we use BERT as an embedding layer followed by a shallow CNN (tri-gram kernel) (Kim, 2014).
|