init_test

app.py

@@ -273,6 +273,7 @@ def extract_face_from_fixed_num_frames(src_video, dst_path, num_frames=None, dev
 
     video_capture.release()
     # cv2.destroyAllWindows()
+    return frame_index
 
 
 def FSFM3C_video_detection(video):
@@ -286,7 +287,7 @@ def FSFM3C_video_detection(video):
     frame_path = os.path.join(FRAME_SAVE_PATH, str(num_files))
     os.makedirs(frame_path, exist_ok=True)
     os.makedirs(os.path.join(frame_path, '0'), exist_ok=True)
-    extract_face_from_fixed_num_frames(video, frame_path, num_frames=num_frames, device=device)
+    frame_index = extract_face_from_fixed_num_frames(video, frame_path, num_frames=num_frames, device=device)
 
     args.data_path = frame_path
     args.batch_size = 32
@@ -300,9 +301,10 @@ def FSFM3C_video_detection(video):
         drop_last=False
     )
 
-    frame_preds_list, video_y_pred_list = test_all(data_loader_val, model, device)
+    frame_preds_list, video_pred_list = test_all(data_loader_val, model, device)
 
-    return video_y_pred_list
+
+    return video_pred_list
 
 
 def FSFM3C_image_detection(image):
@@ -333,9 +335,9 @@ def FSFM3C_image_detection(image):
         drop_last=False
     )
 
-    frame_preds_list, video_y_pred_list = test_all(data_loader_val, model, device)
+    frame_preds_list, video_pred_list = test_all(data_loader_val, model, device)
 
-    return video_y_pred_list
+    return video_pred_list
 
 
 # WebUI
@@ -348,11 +350,11 @@ with gr.Blocks() as demo:
 
     gr.Markdown("- <b>V1.0 [2024-12] (Current):</b> "
                 "Create this page with basic detectors (simply fine-tuned models) that follow the paper implementation. "
-                "<b>Notes:</b> Performance is limited because no any optimization of data, models, hyperparameters, etc. is done for downstream tasks. <br>"
-                "- </b>(TODO):</b> Update practical models, and optimized interfaces, and provide more functions such as visualizations, a unified detector, and multi-modal diagnosis.")
+                "<b>Notes:</b> Performance is limited because no any optimization of data, models, hyperparameters, etc. is done for downstream tasks."
+                "- </b>[TODO]: </b> Update practical models, and optimized interfaces, and provide more functions such as visualizations, a unified detector, and multi-modal diagnosis.")
 
     gr.Markdown(
-        "> Please provide an <b>image</b> or a <b>video (<100s </b>, default to uniform sampling 32 frames)</b> and </b>select the model</b> for detection. <br>"
+        "> Please provide an <b>image</b> or a <b>video (<100s </b>, default to uniform sampling 32 frames)</b> and <b>select the model</b> for detection. <br>"
         "- <b>DfD-Checkpoint_Fine-tuned_on_FF++</b> for deepfake detection, FSFM VIT-B fine-tuned on the FF++_c23 dataset (train&val sets of  4 manipulations, 32 frames per video) <br>"
         "- <b>FAS-Checkpoint_Fine-tuned_on_MCIO</b> for face anti-spoofing, FSFM VIT-B fine-tuned on the MCIO datasets (2 frames per video) ")
 

engine_finetune.py

@@ -320,4 +320,4 @@ def test_all(data_loader, model, device):
     #               video_auc=metric_logger.video_auc, video_eer=metric_logger.video_eer))
 
     # return {k: meter.global_avg for k, meter in metric_logger.meters.items()}
-    return frame_preds_list, video_y_pred_list
+    return frame_preds_list, video_pred_list