Upload folder using huggingface_hub

.gitattributes

@@ -57,3 +57,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 # Video files - compressed
 *.mp4 filter=lfs diff=lfs merge=lfs -text
 *.webm filter=lfs diff=lfs merge=lfs -text
+LongTVQA_plus_train.json filter=lfs diff=lfs merge=lfs -text

.vscode/settings.json

@@ -0,0 +1,5 @@
+{
+    "python-envs.defaultEnvManager": "ms-python.python:conda",
+    "python-envs.defaultPackageManager": "ms-python.python:conda",
+    "python-envs.pythonProjects": []
+}

LongTVQA_plus_train.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2655b7e2954918022564847248887577e20fa8fe871ac9b74f9438ce293367c6
+size 57356136

README.md

@@ -1,3 +1,90 @@
----
-license: apache-2.0
----
+# LongTVQA+ Dataset
+
+This repository contains the **LongTVQA+** dataset in JSON format. LongTVQA+ differs from the original TVQA dataset in three ways:
+
+1. The questions in TVQA+ are a subset of TVQA (The Big Bang Theory).
+2. TVQA+ provides frame-level bounding box annotations for visual concept words in questions and correct answers.
+3. TVQA+ has refined timestamp annotations.
+
+Please refer to the TVQA+ paper for more details.
+
+## Files
+
+- `LongTVQA_plus_train.json` — training split (23,545 QA samples)
+- `LongTVQA_plus_val.json` — validation split (3,017 QA samples)
+- `LongTVQA_plus_subtitles.json` — subtitles indexed by video clip (4,198 clips)
+
+## QA JSON format
+
+Each entry in `LongTVQA_plus_train.json` and `LongTVQA_plus_val.json` is a dictionary with the following fields:
+
+| Key | Type | Description |
+| --- | --- | --- |
+| `qid` | int | Question ID (same as in TVQA). |
+| `q` | str | Question text. |
+| `a0` ... `a4` | str | Five multiple-choice answers. |
+| `answer_idx` | str | Correct answer index (`"0"`-`"4"`). |
+| `ts` | list | Refined timestamp annotation, e.g. `[0, 5.4]` indicates the localized span starts at 0s and ends at 5.4s. |
+| `vid_name` | str | Video clip name. Format: `{show_name_abbr}_s{season}e{episode}_seg{segment}_clip_{clip}`; e.g. `friends_s06e12_seg02_clip_16`. Episodes typically have two segments split by the opening song. Clips for **The Big Bang Theory** omit `{show_name_abbr}` (e.g. `s05e02_seg02_clip_00`). |
+| `bbox` | dict | Bounding boxes for annotated frames (3 FPS). Keys are frame numbers. Values are lists of boxes with `img_id`, `top`, `left`, `width`, `height`, and `label`. |
+
+### QA sample
+
+```json
+{
+  "answer_idx": "1",
+  "qid": 134094,
+  "ts": [5.99, 11.98],
+  "a1": "Howard is talking to Raj and Leonard",
+  "a0": "Howard is talking to Bernadette",
+  "a3": "Howard is talking to Leonard and Penny",
+  "a2": "Howard is talking to Sheldon , and Raj",
+  "q": "Who is Howard talking to when he is in the lab room ?",
+  "vid_name": "s05e02_seg02_clip_00",
+  "a4": "Howard is talking to Penny and Bernadette",
+  "bbox": {
+    "14": [
+      {
+        "img_id": 14,
+        "top": 153,
+        "label": "Howard",
+        "width": 180,
+        "height": 207,
+        "left": 339
+      },
+      {
+        "img_id": 14,
+        "top": 6,
+        "label": "lab",
+        "width": 637,
+        "height": 354,
+        "left": 3
+      }
+    ],
+    "20": [],
+    "26": [],
+    "32": [],
+    "38": []
+  }
+}
+```
+
+## Subtitles JSON format
+
+`LongTVQA_plus_subtitles.json` is a dictionary keyed by `vid_name`. Each value has:
+
+| Key | Type | Description |
+| --- | --- | --- |
+| `sub_text` | str | Subtitle text for the clip, with utterances separated by `<eos>`. |
+| `sub_time` | list | Start times (seconds) corresponding to each subtitle segment. |
+
+### Subtitles sample
+
+```json
+{
+  "s09e14_seg02_clip_04": {
+    "sub_text": "Sheldon : That  's a risk I  'm willing to take ! <eos> Amy : Well , this is so nice . <eos> ...",
+    "sub_time": [0.9550000000000001, 3.825, 5.359, 7.295]
+  }
+}
+```

up.py

@@ -0,0 +1,7 @@
+from huggingface_hub import login, upload_folder
+
+# (optional) Login with your Hugging Face credentials
+login()
+
+# Push your dataset files
+upload_folder(folder_path=".", repo_id="longvideoagent/LongTVQA_plus", repo_type="dataset")