feat: Add Image to Text

controllers/img2text.py

@@ -0,0 +1,15 @@
+from PIL import Image
+from controllers.utils import image_to_bytes
+from services.api_service import post_data
+from .fw import get_fw_query_params
+
+
+def img2text_controller(image: Image, fw_option: str = 'none'):
+    request_data_files = {
+        "image": image_to_bytes(image),
+    }
+    response = post_data(
+        f"/api/img2text/{get_fw_query_params(fw_option)}", files=request_data_files)
+    if response is None:
+        return None
+    return response['caption']

controllers/vqa.py

@@ -13,7 +13,6 @@ def vqa_controller(image: Image, question: str, fw_option: str = 'none'):
     }
     response = post_data(
         f"/api/vqa/{get_fw_query_params(fw_option)}", files=request_data_files, data=request_data_form)
-    print(response)
     if response is None:
         return None
     return response['answer']

views/__init__.py

@@ -3,6 +3,7 @@ import gradio as gr
 from .rembg import rembg_interface
 from .fd import fd_interface
 from .chat import chat_interface
+from .img2text import img2text_interface
 from .vqa import vqa_interface
 from .config import config_interface
 
@@ -11,12 +12,14 @@ def create_interface():
     rembg_ui = rembg_interface()
     fd_ui = fd_interface()
     chat_ui = chat_interface()
-    config_ui = config_interface()
+    img2text_ui = img2text_interface()
     vqa_ui = vqa_interface()
+    config_ui = config_interface()
 
     return gr.TabbedInterface(
         title="GDSC AI Service Inference",
-        interface_list=[rembg_ui, fd_ui, chat_ui.queue(), vqa_ui, config_ui],
+        interface_list=[rembg_ui, fd_ui,
+                        chat_ui.queue(), img2text_ui, vqa_ui, config_ui],
         tab_names=[rembg_ui.title, fd_ui.title,
-                   chat_ui.title, vqa_ui.title, config_ui.title],
+                   chat_ui.title, img2text_ui.title, vqa_ui.title, config_ui.title],
     )

views/img2text.py

@@ -0,0 +1,19 @@
+import gradio as gr
+from services.state import get_state
+from controllers.img2text import img2text_controller
+from utilities.constants import FW_DEFAULT_OPTION
+
+
+def img2text_interface():
+    return gr.Interface(
+        title="Image to Text",
+        fn=lambda image: img2text_controller(
+            image, get_state('fw_option', FW_DEFAULT_OPTION)),
+        inputs=[
+            gr.Image(type="pil", label="Input Image"),
+        ],
+        outputs=[
+            gr.Textbox(label="Caption")
+        ],
+        flagging_options=[]
+    )