feat: Image to Text

DOCS.md → .github/README.md

@@ -72,13 +72,23 @@ This route checks the health status of children server. The children is consider
 ```json
 // Request. Form-data
 {
-    "image": "<image-data"
+    "image": "<image-data>"
 }
 
 // Response. JSON
 {
     "data": {
-        "faces": []
+        "faces": FaceObject[]
+    }
+}
+
+// FaceObject
+{
+    "bbox": {
+        "x": 0,
+        "y": 0,
+        "width": 100,
+        "height": 100
     }
 }
 ```
@@ -125,6 +135,22 @@ This route checks the health status of children server. The children is consider
 }
 ```
 
+- `/api/img2text/`
+
+```json
+// Request. Form-data
+{
+    "image": <image-data>
+}
+
+// Response. JSON
+{
+    "data": {
+        "caption": "Caption of image"
+    }
+}
+```
+
 ## 😊 Contributors
 
 - GDSC-FPTU [[gdsc-fptu](https://github.com/gdsc-fptu)]

Dockerfile

@@ -2,7 +2,7 @@
 FROM python:3.11.7-slim-bookworm
 
 # Set environment variables
-ENV HOME=/home/user \
+ENV CLOUD_HOME=/home/user \
     PATH=/home/user/.local/bin:$PATH
 
 # Install libgl1-mesa-glx for opencv
@@ -16,7 +16,7 @@ RUN apt-get install 'ffmpeg'\
 RUN useradd -m -u 1000 user
 
 # Define working directory
-WORKDIR $HOME/app
+WORKDIR $CLOUD_HOME/app
 
 # Switch to user
 USER user
@@ -29,7 +29,7 @@ RUN pip install --no-cache-dir --upgrade pip
 RUN pip install --user -r /app/requirements.txt
 
 # Copy the rest of the code to the image
-COPY --chown=user:user . $HOME/app
+COPY --chown=user:user . $CLOUD_HOME/app
 
 # Expose port 7860
 EXPOSE 7860/tcp

apps/apis/chats/controllers/openai_controller.py

@@ -7,10 +7,12 @@ BASE_ENHANCE_PROMPT = {
     "role": "system", "content": "You are a helpful assistant. You are helping a customer to solve a problem."}
 
 
-client = OpenAI(api_key=os.getenv("OPENAI_API_KEY"))
+client = OpenAI()
 
 
-def send_message(prompt: str, histories: list = []):
+def send_message(prompt: str, histories: list = [], api_key: str = None):
+    if api_key is not None:
+        client.api_key = api_key
     return client.chat.completions.create(
         model="gpt-3.5-turbo",
         messages=[
@@ -21,7 +23,9 @@ def send_message(prompt: str, histories: list = []):
     ).choices[0].message.content
 
 
-def send_message_conversation(conversation_id: str, prompt: str):
+def send_message_conversation(conversation_id: str, prompt: str, api_key: str = None):
+    if api_key is not None:
+        client.api_key = api_key
     # Update conversation
     update_conversation_storage(conversation_id, "user", prompt)
     # Generate response

apps/apis/chats/models/chats_model.py

@@ -5,3 +5,5 @@ class ChatsModel(BaseModel):
     prompt: str = Field(..., example="Hello, I'm a chatbot")
     histories: list[dict[str, str]] | None = Field(
         None, example=[{"user": "Hello", "assistant": "Hi"}])
+    openai_key: str | None = Field(
+        None, descriptions="Open AI API key", example="OPENAI_KEY")

apps/apis/img2text/__init__.py

@@ -1,5 +1,7 @@
-from fastapi import APIRouter
+from fastapi import APIRouter, Depends
 from fastapi.responses import JSONResponse
+from apps.services.foward.fw_middleware import forward_middleware, forward_request
+from .models.img2text_model import Img2TextModel
 
 router = APIRouter(prefix='/img2text')
 router_base_configs = {
@@ -10,5 +12,16 @@ router_base_configs = {
 
 # Image to text
 @router.post("/", **router_base_configs)
-def image_to_text():
-    return {"caption": "Hello World"}
+def image_to_text(
+    image: Img2TextModel.image = Img2TextModel.image_default,
+    fw_index: Img2TextModel.fw_index = Depends(forward_middleware)
+):
+    # Forward request
+    fw_data = {
+        "files": {"image": image.file}
+    }
+    fw_response = forward_request(fw_index, fw_data, '/api/img2text/')
+    if fw_response is not None:
+        return {"caption": fw_response["data"]["caption"]}
+
+    return {"caption": "Image to Text model is not available in this server"}

apps/apis/img2text/models/img2text_model.py

@@ -0,0 +1,7 @@
+from fastapi import File, UploadFile
+
+
+class Img2TextModel:
+    image = UploadFile | None
+    image_default = File(None, description="Image file")
+    fw_index = int | None