Commit
·
2ceca78
1
Parent(s):
6e7198a
Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,11 @@
|
|
| 1 |
---
|
| 2 |
license: cc-by-4.0
|
| 3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
license: cc-by-4.0
|
| 3 |
---
|
| 4 |
+
This model uses the Deep Fashion dataset in order to create a category classifier among the 50 or so provided categories.
|
| 5 |
+
https://mmlab.ie.cuhk.edu.hk/projects/DeepFashion.html
|
| 6 |
+
|
| 7 |
+
This model leverages the ViT (Vision transformer), loaded with the custom dataset and the 50 odd categoes to which they are assigned. The objective here, is to expand the same and get to
|
| 8 |
+
a. An accuracy level of 90+ in the top 5 categorizes
|
| 9 |
+
b. An accuracy of 70+ overall.
|
| 10 |
+
|
| 11 |
+
In addition, we would also look forward to creating attribute extractors, to extract key attributes (primary color, checked, sleeve, collar etc) as we proceed
|