Summary

app/app.py

@@ -65,16 +65,15 @@ APP = """
         <h3>Journals</h3>
         <pre>
             <code>
-@article{ryumina22_neurocomputing,
-    author = {Elena Ryumina and Denis Dresvyanskiy and Alexey Karpov},
-    title = {In Search of a Robust Facial Expressions Recognition Model: A Large-Scale Visual Cross-Corpus Study},
-    journal = {Neurocomputing},
-    volume = {514},
-    pages = {435-450},
-    year = {2022},
-    doi = {<a href="https://doi.org/10.1016/j.neucom.2022.10.013">https://doi.org/10.1016/j.neucom.2022.10.013</a>},
+@article{ryumina24_prl,
+    author = {Ryumina, Elena and Markitantov, Maxim and Ryumin, Dmitry and Karpov, Alexey},
+    title = {Gated Siamese Fusion Network based on Multimodal Deep and Hand-Crafted Features for Personality Traits Assessment},
+    journal = {Pattern Recognition Letters},
+    volume = {185},
+    pages = {45--51},
+    year = {2024},
+    doi = {<a href="https://doi.org/10.1016/j.patrec.2024.07.004">https://doi.org/10.1016/j.patrec.2024.07.004</a>},
 }
-
 @article{ryumina24_eswa,
     author = {Elena Ryumina and Maxim Markitantov and Dmitry Ryumin and Alexey Karpov},
     title = {OCEAN-AI Framework with EmoFormer Cross-Hemiface Attention Approach for Personality Traits Assessment},
@@ -83,6 +82,15 @@ APP = """
     pages = {122441},
     year = {2024},
     doi = {<a href="https://doi.org/10.1016/j.eswa.2023.122441">https://doi.org/10.1016/j.eswa.2023.122441</a>},
+}
+@article{ryumina22_neurocomputing,
+    author = {Elena Ryumina and Denis Dresvyanskiy and Alexey Karpov},
+    title = {In Search of a Robust Facial Expressions Recognition Model: A Large-Scale Visual Cross-Corpus Study},
+    journal = {Neurocomputing},
+    volume = {514},
+    pages = {435-450},
+    year = {2022},
+    doi = {<a href="https://doi.org/10.1016/j.neucom.2022.10.013">https://doi.org/10.1016/j.neucom.2022.10.013</a>},
 }
             </code>
         </pre>
@@ -90,6 +98,13 @@ APP = """
         <h3>Conferences</h3>
         <pre>
             <code>
+@inproceedings{ryumina24_interspeech,
+    author = {Elena Ryumina and Dmitry Ryumin and and Alexey Karpov},
+    title = {OCEAN-AI: Open Multimodal Framework for Personality Traits Assessment and HR-Processes Automatization},
+    year = {2024},
+    booktitle = {INTERSPEECH},
+    pages = {in press},
+}
 @inproceedings{ryumina23_interspeech,
     author = {Elena Ryumina and Dmitry Ryumin and Maxim Markitantov and Heysem Kaya and Alexey Karpov},
     title = {Multimodal Personality Traits Assessment (MuPTA) Corpus: The Impact of Spontaneous and Read Speech},

app/description.py

@@ -15,6 +15,10 @@ TEMPLATE = """\
     <img src="https://img.shields.io/badge/version-v{version}-rc0" alt="{version_label}">
     <a href='https://github.com/DmitryRyumin/OCEANAI' target='_blank'><img src='https://img.shields.io/github/stars/DmitryRyumin/OCEANAI?style=flat' alt='GitHub' /></a>
 </div>
+
+The models used in OCEAN-AI were trained on 15-second clips from the ChaLearn First Impression v2 dataset. 
+For more reliable predictions, 15-second videos are recommended, but OCEAN-AI can analyze videos of any length. 
+Due to limited computational resources on HuggingFace, we provide six 3-second videos as examples.
 """
 
 DESCRIPTIONS = [

app/event_handlers/practical_subtasks.py

@@ -317,10 +317,21 @@ def event_handler_practical_subtasks(
         or practical_subtasks.lower() == "mobile device application categories"
         or practical_subtasks.lower() == "clothing style correlation"
     ):
-        df_correlation_coefficients = read_csv_file(
-            config_data.Links_CAR_CHARACTERISTICS,
-            ["Trait", "Style and performance", "Safety and practicality"],
-        )
+        if practical_subtasks.lower() == "car characteristics":
+
+            df_correlation_coefficients = read_csv_file(
+                config_data.Links_CAR_CHARACTERISTICS,
+                ["Trait", "Style and performance", "Safety and practicality"],
+            )
+
+        elif practical_subtasks.lower() == "mobile device application categories":
+
+            df_correlation_coefficients = read_csv_file(
+                config_data.Links_MDA_CATEGORIES
+            )
+
+        elif practical_subtasks.lower() == "clothing style correlation":
+            df_correlation_coefficients = read_csv_file(config_data.Links_CLOTHING_SC)
 
         return (
             practical_subtasks_selected,

app/oceanai_init.py

@@ -26,10 +26,10 @@ def oceanai_initialization():
 
     # Загрузка весов аудиомоделей
     url = _b5.weights_for_big5_["audio"][corpus]["hc"]["sberdisk"]
-    _ = _b5.load_audio_model_weights_hc(url=url, out=out)
+    _ = _b5.load_audio_model_weights_hc(url=url, out=out, force_reload=False)
 
     url = _b5.weights_for_big5_["audio"][corpus]["nn"]["sberdisk"]
-    _ = _b5.load_audio_model_weights_nn(url=url, out=out)
+    _ = _b5.load_audio_model_weights_nn(url=url, out=out, force_reload=False)
 
     # Формирование видеомоделей
     _ = _b5.load_video_model_hc(lang="en", out=out)
@@ -38,16 +38,16 @@ def oceanai_initialization():
 
     # Загрузка весов видеомоделей
     url = _b5.weights_for_big5_["video"][corpus]["hc"]["sberdisk"]
-    _ = _b5.load_video_model_weights_hc(url=url, out=out)
+    _ = _b5.load_video_model_weights_hc(url=url, out=out, force_reload=False)
 
     url = _b5.weights_for_big5_["video"][corpus]["fe"]["sberdisk"]
-    _ = _b5.load_video_model_weights_deep_fe(url=url, out=out)
+    _ = _b5.load_video_model_weights_deep_fe(url=url, out=out, force_reload=False)
 
     url = _b5.weights_for_big5_["video"][corpus]["nn"]["sberdisk"]
-    _ = _b5.load_video_model_weights_nn(url=url, out=out)
+    _ = _b5.load_video_model_weights_nn(url=url, out=out, force_reload=False)
 
     # Загрузка словаря с экспертными признаками (текстовая модальность)
-    _ = _b5.load_text_features(out=out)
+    _ = _b5.load_text_features(out=out, force_reload=False)
 
     # Формирование текстовых моделей
     _ = _b5.setup_translation_model()  # только для русского языка
@@ -57,17 +57,17 @@ def oceanai_initialization():
 
     # Загрузка весов текстовых моделей
     url = _b5.weights_for_big5_["text"][corpus]["hc"]["sberdisk"]
-    _ = _b5.load_text_model_weights_hc(url=url, out=out)
+    _ = _b5.load_text_model_weights_hc(url=url, out=out, force_reload=False)
 
     url = _b5.weights_for_big5_["text"][corpus]["nn"]["sberdisk"]
-    _ = _b5.load_text_model_weights_nn(url=url, out=out)
+    _ = _b5.load_text_model_weights_nn(url=url, out=out, force_reload=False)
 
     # Формирование модели для мультимодального объединения информации
     _ = _b5.load_avt_model_b5(out=out)
 
     # Загрузка весов модели для мультимодального объединения информации
     url = _b5.weights_for_big5_["avt"][corpus]["b5"]["sberdisk"]
-    _ = _b5.load_avt_model_weights_b5(url=url, out=out)
+    _ = _b5.load_avt_model_weights_b5(url=url, out=out, force_reload=False)
 
     return _b5
 

video_metadata.yaml

@@ -1,14 +1,14 @@
 video_metadata:
-  1_9093a4ca3c0c834.mp4:
-    - Emily
-    - Taylor
-    - emily.taylor@example.com
-    - "+1 (555) 456-7890"
-  2_a6a198e51d073b0.mp4:
+  1_a6a198e51d073b0.mp4:
     - Michael
     - Brown
     - michael.brown@example.com
     - "+1 (555) 234-5678"
+  2_9093a4ca3c0c834.mp4:
+    - Emily
+    - Taylor
+    - emily.taylor@example.com
+    - "+1 (555) 456-7890"
   3_9987232dd677712.mp4:
     - Jack
     - Smith