chore(space): Dockerize FastAPI app; avoid Gradio/xformers on server; add README

Dockerfile

@@ -0,0 +1,32 @@
+# syntax=docker/dockerfile:1
+
+FROM python:3.10-slim
+
+ENV PIP_NO_CACHE_DIR=1 \
+    PYTHONDONTWRITEBYTECODE=1 \
+    PYTHONUNBUFFERED=1 \
+    HF_HUB_ENABLE_HF_TRANSFER=1
+
+WORKDIR /app
+
+# System deps
+RUN apt-get update && apt-get install -y --no-install-recommends \
+    git \
+    gcc \
+    libgl1 \
+    && rm -rf /var/lib/apt/lists/*
+
+# Copy code
+COPY . /app
+
+# Install deps
+RUN python -m pip install --upgrade pip && \
+    python -m pip install -r requirements.txt && \
+    # CPU-only torch to avoid CUDA/xformers issues on Spaces CPU images
+    python -m pip install --index-url https://download.pytorch.org/whl/cpu torch torchvision torchaudio --upgrade --force-reinstall && \
+    # Ensure xformers is not installed
+    python -m pip uninstall -y xformers || true
+
+EXPOSE 7860
+
+CMD ["python", "-m", "uvicorn", "server:app", "--host", "0.0.0.0", "--port", "7860"]

README.md

@@ -12,3 +12,27 @@ short_description: hair stable
 ---
 
 Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+
+# Hair_stable_new
+
+## API (FastAPI)
+
+Base URL (Space): <your Space URL>
+
+Bearer Token: `logicgo@123`
+
+Endpoints:
+- GET /health
+- POST /upload (form-data: image=File) [Bearer]
+- POST /get-hairswap (JSON) [Bearer]
+- GET /download/{filename} [Bearer]
+- GET /logs [Bearer]
+
+Run locally:
+```bash
+python3 -m uvicorn server:app --host 0.0.0.0 --port 7861
+```
+
+On Hugging Face Spaces (Docker):
+- Space type: Docker
+- This repo includes a Dockerfile which starts the FastAPI app on port 7860.

app.py

@@ -1,4 +1,7 @@
-import gradio as gr
+try:
+    import gradio as gr
+except Exception:
+    gr = None
 import torch
 from PIL import Image
 import numpy as np
@@ -161,23 +164,22 @@ def model_call(id_image, ref_hair, converter_scale, scale, guidance_scale, contr
     return id_image_bald, image
 
 # Create a Gradio interface
-iface = gr.Interface(
-    fn=model_call,
-    inputs=[
-        gr.Image(label="ID Image"),
-        gr.Image(label="Reference Hair"),
-        gr.Slider(minimum=0.5, maximum=1.5, value=1, label="Converter Scale"),
-        gr.Slider(minimum=0.0, maximum=3.0, value=1.0, label="Hair Encoder Scale"),
-        gr.Slider(minimum=1.1, maximum=3.0, value=1.5, label="CFG"),
-        gr.Slider(minimum=0.1, maximum=2.0, value=1, label="Latent IdentityNet Scale"),
-    ],
-    outputs=[
-        gr.Image(type="pil", label="Bald Result"),
-        gr.Image(type="pil", label="Transfer Result"),
-    ],
-    title="Hair Transfer Demo",
-    description="In general, aligned faces work well, but can also be used on non-aligned faces, and you need to resize to 512 * 512"
-)
-
-# Launch the Gradio interface
-iface.queue().launch(server_name='0.0.0.0', server_port=7860, share=True)
+if gr is not None:
+    iface = gr.Interface(
+        fn=model_call,
+        inputs=[
+            gr.Image(label="ID Image"),
+            gr.Image(label="Reference Hair"),
+            gr.Slider(minimum=0.5, maximum=1.5, value=1, label="Converter Scale"),
+            gr.Slider(minimum=0.0, maximum=3.0, value=1.0, label="Hair Encoder Scale"),
+            gr.Slider(minimum=1.1, maximum=3.0, value=1.5, label="CFG"),
+            gr.Slider(minimum=0.1, maximum=2.0, value=1, label="Latent IdentityNet Scale"),
+        ],
+        outputs=[
+            gr.Image(type="pil", label="Bald Result"),
+            gr.Image(type="pil", label="Transfer Result"),
+        ],
+        title="Hair Transfer Demo",
+        description="In general, aligned faces work well, but can also be used on non-aligned faces, and you need to resize to 512 * 512"
+    )
+    iface.queue().launch(server_name='0.0.0.0', server_port=7860, share=True)

infer_full.py

@@ -1,4 +1,3 @@
-import gradio as gr
 import torch
 from PIL import Image
 import numpy as np