remove torch depends from frontend

app.py

@@ -6,7 +6,6 @@
 
 import os
 import cv2
-import torch
 import numpy as np
 import gradio as gr
 import sys
@@ -95,7 +94,6 @@ def _get_result(job_id: str, token: str) -> dict:
 # -------------------------------------------------------------------------
 # 1) Core model inference (now forwards to remote service)
 # -------------------------------------------------------------------------
-@spaces.GPU(duration=120)
 def run_model(target_dir, model=None) -> dict:
     """
     Run the VGGT model on images in the 'target_dir/images' folder and return predictions.
@@ -189,7 +187,6 @@ def handle_uploads(input_video, input_images):
     """
     start_time = time.time()
     gc.collect()
-    torch.cuda.empty_cache()
 
     # Create a unique folder name
     timestamp = datetime.now().strftime("%Y%m%d_%H%M%S_%f")
@@ -274,7 +271,6 @@ def update_gallery_on_upload(input_video, input_images):
 # -------------------------------------------------------------------------
 # 4) Reconstruction: uses the target_dir plus any viz parameters
 # -------------------------------------------------------------------------
-@spaces.GPU(duration=120)
 def gradio_demo(
     target_dir,
     conf_thres=3.0,
@@ -293,7 +289,6 @@ def gradio_demo(
 
     start_time = time.time()
     gc.collect()
-    torch.cuda.empty_cache()
 
     # Prepare frame_filter dropdown
     target_dir_images = os.path.join(target_dir, "images")
@@ -306,8 +301,7 @@ def gradio_demo(
     frame_filter_choices = ["All"] + all_files
 
     print("Running run_model...")
-    with torch.no_grad():
-        predictions = run_model(target_dir)
+    predictions = run_model(target_dir)
 
     # Save predictions
     prediction_save_path = os.path.join(target_dir, "predictions.npz")
@@ -340,7 +334,6 @@ def gradio_demo(
     # Cleanup
     del predictions
     gc.collect()
-    torch.cuda.empty_cache()
 
     end_time = time.time()
     print(f"Total time: {end_time - start_time:.2f} seconds")