Add data

README.md

@@ -17,4 +17,12 @@ task_ids:
 ---
 # TextCaps in Vietnamese
 
-WIP
+This is Vietnamese version of [TextCaps dataset](https://textvqa.org/textcaps/). It has 109765 image-caption pairs for training, and 15830 ones for validation. It was built by using Google Translate API. The Vietnamese version has the almost metadata as English one. The Vietnamese version doesn't have the following keys for each data points.
+- `caption_tokens`
+- `reference_tokens`
+- `reference_strs`
+- `image_classes`
+
+For English version, these keys are in English. Because my main focus is `caption_str`, there aren't Vietnamese version for them. I'm limited by time and disk space.
+
+I provide both English and Vietnamese .json files.

vi/TextCaps_0.1_train.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:98b505f7e1fdf98b396e12c42b33577e7df70363000b62eb873c5ca5ddd6c958
+size 52756665

vi/TextCaps_0.1_val.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ae601f89db46bce156b92bc81b6e1a716a23df79ff9b1a680648d75fc8e05941
+size 7519741