Upload 2 files

README.md

@@ -13,19 +13,20 @@ startup_duration_timeout: 1h
 
 Docker Space for swapping a face from an uploaded image onto a target video.
 
-## Setup
-
-- Create a new Space with Docker SDK.
-- Add the files from this repo.
-- Set hardware to a paid GPU such as A10G small.
-- Add `HF_TOKEN` if your model repo is gated or private.
-
 ## Environment variables
 
-- `MODEL_REPO_ID`: model repo ID, default `ezioruan/inswapper_128.onnx`
-- `MODEL_FILENAME`: model filename, default `inswapper_128.onnx`
+- `MODEL_REPO_ID`: default `ezioruan/inswapper_128.onnx`
+- `MODEL_FILENAME`: default `inswapper_128.onnx`
 - `MODEL_REVISION`: optional branch, tag, or commit hash
-- `HF_TOKEN`: optional token for gated/private repos
-- `FACE_MODEL_NAME`: InsightFace detector name, default `buffalo_l`
-- `DETECTION_SIZE`: detector input size, default `640`
+- `HF_TOKEN`: recommended for gated/private repos and better Hub rate limits
+- `FACE_MODEL_NAME`: default `buffalo_l`
+- `DETECTION_SIZE`: default `640`
 - `MAX_FRAMES`: optional frame cap for testing, default `0`
+- `FORCE_CPU`: set to `1` to force CPU mode for debugging
+
+## Deploy
+
+1. Create a new Space with Docker SDK.
+2. Add the files from this repo.
+3. Select a paid GPU hardware tier such as A10G small.
+4. Add `HF_TOKEN` if you use a private or rate-limited model repo.

app.py

@@ -6,6 +6,7 @@ from pathlib import Path
 
 import cv2
 import gradio as gr
+import onnxruntime as ort
 from huggingface_hub import hf_hub_download
 from insightface.app import FaceAnalysis
 from insightface.model_zoo import get_model
@@ -23,6 +24,7 @@ HF_TOKEN = os.environ.get("HF_TOKEN") or os.environ.get("HUGGINGFACEHUB_API_TOKE
 FACE_MODEL_NAME = os.environ.get("FACE_MODEL_NAME", "buffalo_l")
 DETECTION_SIZE = int(os.environ.get("DETECTION_SIZE", "640"))
 MAX_FRAMES = int(os.environ.get("MAX_FRAMES", "0"))
+FORCE_CPU = os.environ.get("FORCE_CPU") == "1"
 
 face_analyzer = None
 face_swapper = None
@@ -35,7 +37,16 @@ def ensure_ffmpeg():
 
 
 def get_execution_providers():
-    return ["CUDAExecutionProvider", "CPUExecutionProvider"]
+    providers = []
+    if not FORCE_CPU:
+        providers.append("CUDAExecutionProvider")
+    providers.append("CPUExecutionProvider")
+    return providers
+
+
+def provider_status():
+    available = ort.get_available_providers()
+    return available, ("CUDAExecutionProvider" in available)
 
 
 def ensure_swap_model() -> str:
@@ -97,6 +108,12 @@ def swap_frame(frame_bgr, source_face, analyzer, swapper):
 
 
 def process_video(face_image_path, video_path, progress=gr.Progress(track_tqdm=False)):
+    available_providers, cuda_available = provider_status()
+    if FORCE_CPU:
+        raise RuntimeError(f"FORCE_CPU=1 is set. Available providers: {available_providers}")
+    if not cuda_available:
+        raise RuntimeError(f"CUDAExecutionProvider is not available. Available providers: {available_providers}")
+
     ensure_ffmpeg()
     analyzer, swapper = load_models()
     src = cv2.imread(face_image_path)
@@ -136,12 +153,18 @@ def process_video(face_image_path, video_path, progress=gr.Progress(track_tqdm=F
     cap.release()
     writer.release()
     mux_audio(silent_video, audio_file, final_video)
-    return final_video, f"Done. Processed {idx} frame(s)."
+    return final_video, f"Done. Processed {idx} frame(s). Providers: {available_providers}"
+
+
+def startup_message():
+    available, cuda_ok = provider_status()
+    return f"ONNX Runtime providers: {available} | CUDA available: {cuda_ok}"
 
 
-with gr.Blocks(theme=gr.themes.Soft(), title="Video FaceSwap GPU") as demo:
+with gr.Blocks(title="Video FaceSwap GPU") as demo:
     gr.Markdown("""# Video FaceSwap GPU
 Upload a source face image and a target video. The app downloads the swap model from the Hugging Face Hub at runtime, then swaps the main source face onto detected faces in the video.""")
+    status_banner = gr.Textbox(label="Startup status", value=startup_message(), interactive=False)
     with gr.Row():
         face_image = gr.Image(type="filepath", label="Source face image")
         target_video = gr.Video(label="Target video")
@@ -151,4 +174,4 @@ Upload a source face image and a target video. The app downloads the swap model
     run_btn.click(fn=process_video, inputs=[face_image, target_video], outputs=[output_video, status], api_name="faceswap_video")
 
 if __name__ == "__main__":
-    demo.queue(max_size=8).launch(server_name="0.0.0.0", server_port=7860)
+    demo.queue(max_size=8).launch(server_name="0.0.0.0", server_port=7860, share=False)