changed extra_urls to json format, added pk-hfad-2

sign-language-translator · Nov 4, 2023 · 7c1e21f · 7c1e21f
1 parent 1456b64
commit 7c1e21f
Show file tree

Hide file tree

Showing 6 changed files with 102 additions and 32 deletions.
diff --git a/.gitignore b/.gitignore
@@ -11,7 +11,7 @@ sign_recordings/reference_clips/**/*.*
 sign_recordings/videos/**/*.*
 
 text_corpora/raw_corpora/**/*.*
-text_corpora/supported_substrings_frequency.json
+text_corpora/supported_substrings_frequency*.json
 
 # keep these so that directory is visible at least
 !sign_recordings/features/**/.gitkeep

diff --git a/README.md b/README.md
@@ -1,9 +1,11 @@
 # sign-language-datasets
-Datasets for sign language translation. [See release files for big data files]()
+
+Datasets for sign language translation. [See release files for big data files](https://github.com/sign-language-translator/sign-language-datasets/releases)
 
 This project includes:
+
 - Reference Clips
-  - pakistan-hamza_foundation (pk-hfad) [788]
+  - pakistan-hamza_foundation (pk-hfad) [788+1]
   - wordless (~10 minutes of a person doing everything except signs)
 - Word Mapping (which spoken language words map to which sign language clips or sequence of clips)
 - Videos (Recordings)
@@ -13,20 +15,25 @@ This project includes:
 - Text Corpora
   - A parallel corpus of spoken language sentences and their corresponding sign language gloss (clip sequence)
 
-# Definitions
+## Definitions
+
 - gloss: word sequence corresponding to sign sequence for a given complete spoken language text.
 
-# Explanations
+## Explanations
+
 - json ...
 
-# Directory Tree
+## Directory Tree
+
+```text
     sign-language-datasets
     ├── sign_recordings
     │   ├── features
     │   │   └── landmarks
     │   │
     │   ├── reference_clips
     │   │   ├── pk-hfad-1 [788]
+    │   │   ├── pk-hfad-2 [1]
     │   │   └── wordless [1]
     │   │
     │   └── videos
@@ -63,9 +70,12 @@ This project includes:
         ├── supported_substrings_frequency.json
         └── raw_corpora
             └── wikipedia.json
-
-# Bonus
-##### new file size command
 ```
+
+## Bonus
+
+**new file size command**
+
+```bash
 git status -u --porcelain | awk '{print $2}' | xargs ls -hl | awk '{print $5 "\t" $9}'
-```
+```
diff --git a/extra_urls.json b/extra_urls.json
@@ -0,0 +1,60 @@
+{
+  "datasets": {
+    "features": [
+      {
+        "description": "mediapipe_face_v2_hand_v1",
+        "file_to_url": {
+          "features/mediapipe-face-v2-hand-v1_pk-hfad-1_df00001.jsonl.zip": "https://example.com"
+        }
+      }
+    ],
+    "reference_clips": [
+      {
+        "description": "wordless",
+        "file_to_url": {
+          "videos/wordless_wordless.mp4": "https://github.com/sign-language-translator/sign-language-datasets/releases/download/v0.0.2/wordless_wordless.mp4"
+        }
+      },
+      {
+        "description": "pk-hfad-1",
+        "file_to_url": {
+          "videos/pk-hfad-1_1.mp4": "https://github.com/sign-language-translator/sign-language-datasets/releases/download/v0.0.2/pk-hfad-1_1.mp4",
+          "videos/pk-hfad-1_10.mp4": "https://github.com/sign-language-translator/sign-language-datasets/releases/download/v0.0.2/pk-hfad-1_10.mp4",
+          "videos/pk-hfad-1_100.mp4": "https://github.com/sign-language-translator/sign-language-datasets/releases/download/v0.0.2/pk-hfad-1_100.mp4",
+          "videos/pk-hfad-1_1000.mp4": "https://github.com/sign-language-translator/sign-language-datasets/releases/download/v0.0.2/pk-hfad-1_1000.mp4",
+          "videos/pk-hfad-1_10000.mp4": "https://github.com/sign-language-translator/sign-language-datasets/releases/download/v0.0.2/pk-hfad-1_10000.mp4",
+          "videos/pk-hfad-1_100000.mp4": "https://github.com/sign-language-translator/sign-language-datasets/releases/download/v0.0.2/pk-hfad-1_100000.mp4",
+          "videos/pk-hfad-1_10000000.mp4": "https://github.com/sign-language-translator/sign-language-datasets/releases/download/v0.0.2/pk-hfad-1_10000000.mp4",
+          "videos/pk-hfad-1_4.mp4": "https://github.com/sign-language-translator/sign-language-datasets/releases/download/v0.0.2/pk-hfad-1_4.mp4",
+          "videos/pk-hfad-1_5.mp4": "https://github.com/sign-language-translator/sign-language-datasets/releases/download/v0.0.2/pk-hfad-1_5.mp4",
+          "videos/pk-hfad-1_6.mp4": "https://github.com/sign-language-translator/sign-language-datasets/releases/download/v0.0.2/pk-hfad-1_6.mp4",
+          "videos/pk-hfad-1_7.mp4": "https://github.com/sign-language-translator/sign-language-datasets/releases/download/v0.0.2/pk-hfad-1_7.mp4",
+          "videos/pk-hfad-1_8.mp4": "https://github.com/sign-language-translator/sign-language-datasets/releases/download/v0.0.2/pk-hfad-1_8.mp4",
+          "videos/pk-hfad-1_9.mp4": "https://github.com/sign-language-translator/sign-language-datasets/releases/download/v0.0.2/pk-hfad-1_9.mp4",
+          "videos/pk-hfad-1_a(double-handed-letter).mp4": "https://github.com/sign-language-translator/sign-language-datasets/releases/download/v0.0.2/pk-hfad-1_a.double-handed-letter.mp4",
+          "videos/pk-hfad-1_chips(fries).mp4": "https://github.com/sign-language-translator/sign-language-datasets/releases/download/v0.0.2/pk-hfad-1_chips.fries.mp4",
+          "videos/pk-hfad-1_f(single-handed-letter).mp4": "https://github.com/sign-language-translator/sign-language-datasets/releases/download/v0.0.2/pk-hfad-1_f.single-handed-letter.mp4",
+          "videos/pk-hfad-1_february.mp4": "https://github.com/sign-language-translator/sign-language-datasets/releases/download/v0.0.2/pk-hfad-1_february.mp4",
+          "videos/pk-hfad-1_india.mp4": "https://github.com/sign-language-translator/sign-language-datasets/releases/download/v0.0.2/pk-hfad-1_india.mp4",
+          "videos/pk-hfad-1_strawberry.mp4": "https://github.com/sign-language-translator/sign-language-datasets/releases/download/v0.0.2/pk-hfad-1_strawberry.mp4",
+          "videos/pk-hfad-1_washroom.mp4": "https://github.com/sign-language-translator/sign-language-datasets/releases/download/v0.0.2/pk-hfad-1_washroom.mp4"
+        }
+      },
+      {
+        "description": "pk-hfad-2",
+        "file_to_url": {
+          "videos/pk-hfad-2_hour.mp4": "https://github.com/sign-language-translator/sign-language-datasets/releases/download/v0.0.2/pk-hfad-2_hour.mp4"
+        }
+      }
+    ],
+    "video_recordings": [
+      {
+        "description": "pk-hfad-1",
+        "file_to_url": {
+          "videos/pk-hfad-1_df00001_front.zip": "https://example.com"
+        }
+      }
+    ]
+  },
+  "models": {}
+}
diff --git a/extra_urls.yaml b/extra_urls.yaml
diff --git a/sign_recordings/collection_to_label_to_language_to_words.json b/sign_recordings/collection_to_label_to_language_to_words.json
@@ -8543,6 +8543,25 @@
             ]
         }
     },
+    "pk-hfad-2": {
+        "hour": {
+            "english": [
+                "hour"
+            ],
+            "urdu": [
+                "گھنٹہ",
+                "گھنٹے"
+            ],
+            "hindi": [
+                "घंटा(समय)",
+                "घंटे"
+            ],
+            "roman-urdu": [
+                "ghanta",
+                "ghanty"
+            ]
+        }
+    },
     "wordless": {
         "wordless": {
             "english": [

diff --git a/sign_recordings/recordings_labels.json b/sign_recordings/recordings_labels.json
@@ -791,5 +791,8 @@
         "یقینا",
         "یہ",
         "یہاں"
+    ],
+    "pk-hfad-2": [
+        "hour"
     ]
 }