Configure for Hugging Face Spaces CPU compatibility

README.md

@@ -1,3 +1,13 @@
+---
+title: Pixel Prompt Annotator
+emoji: ✨
+colorFrom: blue
+colorTo: green
+sdk: streamlit
+app_file: app.py
+pinned: false
+---
+
 # ✨ Annotation Assistant
 
 ![Demo](demo.jpg)
@@ -32,7 +42,13 @@ Don't just trust the box. The Assistant provides a **Reasoning Stream** explaini
 3.  Add your **Ngrok Authtoken** in the designated cell.
 4.  Run all cells. The app will launch via a public URL.
 
-### 💻 Option 2: Local System (Requires GPU)
+### 🤗 Option 2: Hugging Face Spaces (CPU/GPU)
+1.  Create a new Space on Hugging Face.
+2.  Select **Streamlit** as the SDK.
+3.  Upload the files from this repository.
+4.  The app will build and launch automatically.
+
+### 💻 Option 3: Local System (Requires GPU)
 1.  **Clone the Repo**:
     ```bash
     git clone https://github.com/devsingh02/Pixel-Prompt-Annotator.git

utils.py

@@ -20,12 +20,16 @@ def load_model():
     """
     print(f"Loading model: {MODEL_ID}...")
     try:
+        device_type = "cuda" if torch.cuda.is_available() else "cpu"
+        torch_dtype = torch.float16 if device_type == "cuda" else torch.float32
+        print(f"Using device: {device_type}, dtype: {torch_dtype}")
+
         processor = AutoProcessor.from_pretrained(MODEL_ID, trust_remote_code=True)
         model = AutoModelForVision2Seq.from_pretrained(
             MODEL_ID,
             device_map="auto",
             trust_remote_code=True,
-            torch_dtype=torch.float16
+            torch_dtype=torch_dtype
         )
     except Exception as e:
         print(f"Error loading {MODEL_ID}: {e}")