File size: 5,842 Bytes
61e55d1
 
 
151f388
 
 
 
61e55d1
 
151f388
61e55d1
 
 
 
151f388
 
 
 
61e55d1
 
 
151f388
61e55d1
 
 
 
 
 
151f388
61e55d1
151f388
61e55d1
 
 
151f388
61e55d1
 
 
151f388
 
 
 
 
61e55d1
 
 
151f388
61e55d1
 
 
 
 
 
151f388
61e55d1
151f388
61e55d1
 
 
151f388
61e55d1
 
 
 
151f388
 
 
 
61e55d1
 
 
151f388
61e55d1
 
 
 
 
 
151f388
61e55d1
151f388
61e55d1
 
 
151f388
61e55d1
 
 
 
 
 
 
 
 
 
 
 
151f388
61e55d1
 
 
 
 
151f388
61e55d1
151f388
61e55d1
 
 
151f388
61e55d1
 
 
 
 
 
 
 
 
 
 
 
151f388
61e55d1
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
[
  {
    "test_id": 1,
    "name": "Real-image pipeline test",
    "category": "VALID_IMAGE_PIPELINE",
    "input": "C:\\Users\\user\\Desktop\\multi model connection\\benchmark\\test_images\\real.jpg",
    "ground_truth_label": "UNKNOWN",
    "expected_outcome": "SUCCESSFUL_INFERENCE",
    "operational_status": "SUCCESSFUL_INFERENCE",
    "domain_assessment": "NOT_EVALUATED: ground truth unknown",
    "result": {
      "detector_name": "ViTDeepfakeDetector",
      "model_name": "dima806/deepfake_vs_real_image_detection (ViT-Base/16)",
      "prediction": "REAL",
      "fake_probability": 0.005097,
      "real_probability": 0.994903,
      "raw_logit": -5.273998,
      "confidence": 0.994903,
      "threshold": 0.5,
      "device": "cpu",
      "preprocessing": "ViTImageProcessor(Resize 224x224, Rescale 1/255, Normalize mean=[0.5, 0.5, 0.5], std=[0.5, 0.5, 0.5])",
      "inference_time_ms": 8288.12,
      "error": null,
      "warning": "Vision Transformer trained for facial deepfake detection. Evaluates learned visual artifact representations; does not guarantee authenticity."
    }
  },
  {
    "test_id": 2,
    "name": "AI-generated image",
    "category": "VALID_AI_IMAGE",
    "input": "C:\\Users\\user\\Desktop\\multi model connection\\benchmark\\test_images\\ai_generated.jpg",
    "ground_truth_label": "FAKE",
    "expected_outcome": "SUCCESSFUL_INFERENCE",
    "operational_status": "SUCCESSFUL_INFERENCE",
    "domain_assessment": "MATCH",
    "result": {
      "detector_name": "ViTDeepfakeDetector",
      "model_name": "dima806/deepfake_vs_real_image_detection (ViT-Base/16)",
      "prediction": "FAKE",
      "fake_probability": 0.511827,
      "real_probability": 0.488173,
      "raw_logit": 0.047315,
      "confidence": 0.511827,
      "threshold": 0.5,
      "device": "cpu",
      "preprocessing": "ViTImageProcessor(Resize 224x224, Rescale 1/255, Normalize mean=[0.5, 0.5, 0.5], std=[0.5, 0.5, 0.5])",
      "inference_time_ms": 673.04,
      "error": null,
      "warning": "Vision Transformer trained for facial deepfake detection. Evaluates learned visual artifact representations; does not guarantee authenticity."
    }
  },
  {
    "test_id": 3,
    "name": "Manipulated-style image",
    "category": "VALID_MANIPULATED_IMAGE",
    "input": "C:\\Users\\user\\Desktop\\multi model connection\\benchmark\\test_images\\manipulated.jpg",
    "ground_truth_label": "MANIPULATED",
    "expected_outcome": "SUCCESSFUL_INFERENCE",
    "operational_status": "SUCCESSFUL_INFERENCE",
    "domain_assessment": "NOT_EVALUATED: ground truth is not a supported binary REAL/FAKE label",
    "result": {
      "detector_name": "ViTDeepfakeDetector",
      "model_name": "dima806/deepfake_vs_real_image_detection (ViT-Base/16)",
      "prediction": "REAL",
      "fake_probability": 0.003638,
      "real_probability": 0.996362,
      "raw_logit": -5.612666,
      "confidence": 0.996362,
      "threshold": 0.5,
      "device": "cpu",
      "preprocessing": "ViTImageProcessor(Resize 224x224, Rescale 1/255, Normalize mean=[0.5, 0.5, 0.5], std=[0.5, 0.5, 0.5])",
      "inference_time_ms": 793.58,
      "error": null,
      "warning": "Vision Transformer trained for facial deepfake detection. Evaluates learned visual artifact representations; does not guarantee authenticity."
    }
  },
  {
    "test_id": 4,
    "name": "Corrupted image file",
    "category": "CORRUPTED_FILE",
    "input": "C:\\Users\\user\\Desktop\\multi model connection\\benchmark\\test_images\\corrupted_image.png",
    "ground_truth_label": "CORRUPTED",
    "expected_outcome": "INPUT_REJECTION",
    "operational_status": "INPUT_REJECTION_VERIFIED",
    "domain_assessment": "NOT_EVALUATED: inference/input error",
    "result": {
      "detector_name": "ViTDeepfakeDetector",
      "model_name": "dima806/deepfake_vs_real_image_detection (ViT-Base/16)",
      "prediction": "ERROR",
      "fake_probability": null,
      "real_probability": null,
      "raw_logit": null,
      "confidence": null,
      "threshold": null,
      "device": "cpu",
      "preprocessing": "ViTImageProcessor(Resize 224x224, Rescale 1/255, Normalize mean=[0.5, 0.5, 0.5], std=[0.5, 0.5, 0.5])",
      "inference_time_ms": null,
      "error": "Invalid or corrupted image: cannot identify image file 'C:\\\\Users\\\\user\\\\Desktop\\\\multi model connection\\\\benchmark\\\\test_images\\\\corrupted_image.png'",
      "warning": "Vision Transformer trained for facial deepfake detection. Evaluates learned visual artifact representations; does not guarantee authenticity."
    }
  },
  {
    "test_id": 5,
    "name": "Plain text submitted as image",
    "category": "UNSUPPORTED_INPUT",
    "input": "C:\\Users\\user\\Desktop\\multi model connection\\benchmark\\test_images\\_temp_vit_invalid.txt",
    "ground_truth_label": "INVALID_TYPE",
    "expected_outcome": "INPUT_REJECTION",
    "operational_status": "INPUT_REJECTION_VERIFIED",
    "domain_assessment": "NOT_EVALUATED: inference/input error",
    "result": {
      "detector_name": "ViTDeepfakeDetector",
      "model_name": "dima806/deepfake_vs_real_image_detection (ViT-Base/16)",
      "prediction": "ERROR",
      "fake_probability": null,
      "real_probability": null,
      "raw_logit": null,
      "confidence": null,
      "threshold": null,
      "device": "cpu",
      "preprocessing": "ViTImageProcessor(Resize 224x224, Rescale 1/255, Normalize mean=[0.5, 0.5, 0.5], std=[0.5, 0.5, 0.5])",
      "inference_time_ms": null,
      "error": "Invalid or corrupted image: cannot identify image file 'C:\\\\Users\\\\user\\\\Desktop\\\\multi model connection\\\\benchmark\\\\test_images\\\\_temp_vit_invalid.txt'",
      "warning": "Vision Transformer trained for facial deepfake detection. Evaluates learned visual artifact representations; does not guarantee authenticity."
    }
  }
]