Anomalies are local events, yet weakly supervised video anomaly detection scores them globally. Most methods pool an entire frame into a single vector, rank frames by that vector, and derive a location afterwards as a saliency map of the score. Full-frame pooling mixes the anomaly with its background, so a detector can reach competitive benchmark scores from scene context without being able to show where in the scene the anomaly occurs. We introduce Region-VAD, which first selects a region and then detects the anomaly from that region alone. Region-VAD selects an explicit rectangle over the tokens of a frozen foundation model and scores anomalousness only from the tokens inside the box, through attention pooling followed by a standard temporal head. The rectangle is not learned, it is the box in which the frozen tokens change most between consecutive segments, in all four feature projections of the encoder, and it is found by exhaustive search over the rectangles of the token grid. Only the pooling and the temporal head are trained, from video-level labels without boxes, masks or frame-level annotations, and no trained component reads a token outside the selected region. Experiments on five datasets, with one localizer configuration shared by all datasets and backbones, show that Region-VAD achieves state-of-the-art anomaly localization on UCF-Crime and MSAD at competitive detection performance.
Figures & tables
Figure 2: Overview of Region-VAD. \raisebox{-0.6pt}{\scriptsize1}⃝ A frozen C-RADIOv4 encoder maps the middle frame of each 16-frame segment to a 20×15 token grid in four projections. \raisebox{-0.6pt}{\scriptsize2}⃝ Region selection has no learnable parameter. The standardized token change from segment t−1 to t gives one map per projection. Their per-patch minimum, the consensus mt , is averaged over t−1 , t , t+1 into m~t , and Rt is the rectangle with the largest centred sum of m~t . \raisebox{-0.6pt}{\scriptsize3}⃝ The tokens inside Rt are pooled by two trained attention stages, across the four projections into one map of d -dimensional tokens and then across the patches into ft∈Rd , which a VAD head scores. The video label supervises only the blocks marked with a flame. Fields and boxes are outputs of the method on UCF-Crime RoadAccidents121.
Figure 3: Change maps of one frame. UCF-Crime Fighting042, with the ground truth of [ 23 ] dashed in green. The change ctk from segment t−1 to t in the four projections of C-RADIOv4, standardized and on one colour scale, and their consensus mt (white: the rectangle selected from each projection’s map alone; orange: Rt , selected from the temporal average m~t ). Each projection marks patches that the others miss, and the consensus keeps the patches marked by all four.
area (% of the frame)
centre distance (px)
annotated
Rt (ours)
annot.
Rt
Annotation set
mean
med.
mean
med.
mean
mean
UCF-Crime [ 23 ]
23.1
20.1
29.5
27.0
52.8
54.1
UCF-Crime [ 1 ]
28.4
23.5
30.6
28.0
58.8
55.3
MSAD [ 1 ]
28.0
19.4
27.2
25.7
61.8
61.1
XD-Violence [ 1 ]
69.1
76.2
47.2
48.0
22.0
36.6
Table 1: Size and position of the boxes. Mean and median area (% of the frame) of the annotated boxes and of Rt over the annotated anomalous test frames, and mean Euclidean distance (pixels of the 320×240 frame) from the box centre to the frame centre.
UCF-Crime
MSAD
XD-Violence
UBnormal
ShanghaiTech
Method
Backbone
TIoU [ 23 ]
TIoU [ 1 ]
AUC
TIoU [ 1 ]
AUC
TIoU [ 1 ]
AP
TIoU
AUC
TIoU
AUC
Controls
Full frame
C-RADIOv4
23.07
28.42
85.94
27.54
87.98
68.98
71.44
21.65
63.52
3.85
96.96
Centred box
C-RADIOv4
34.02
30.94
84.84
26.33
89.03
50.21
75.03
20.36
59.88
6.20
95.61
Fixed 8×8 -patch box
C-RADIOv4
30.76
27.08
85.48
22.48
88.39
26.66
71.70
21.62
64.01
5.94
96.55
Frame-difference box
pixels
45.96
41.25
85.65
40.57
89.55
44.76
76.83
48.71
67.45
12.92
94.90
Table 2: State-of-the-art comparison on five benchmarks. Localization and detection of the methods with a spatial output and of four controls. Three box controls test whether a box scores by its placement or by its size, and the frame-difference box applies the rule of Sec. 3.2 to pixel differences. For detection, the same head scores the mean of each control’s tokens (Sec. 4.1 ). TIoU [ 23 ] : boxes of [ 23 ] on every 16th frame; TIoU [ 1 ] : dense boxes of [ 1 ] ; UBnormal and ShanghaiTech: official masks; † : unreleased keyframe boxes of [ 16 ] , not comparable with ours. Numbers of published methods are those reported in [ 16 ] , [ 1 ] and the respective papers. Bold and underline mark the best and second-best per column, including the controls, and ties share the mark. Last four rows: gain of Region-VAD over the four controls.
token-change maps
UCF-Crime
backbone
DINOv3
SAM 3
SigLIP 2
smoothing
TIoU [ 23 ]
TIoU [ 1 ]
✓
✗
✗
✗
✓
42.75
40.71
✗
✓
✗
✗
✓
45.60
42.45
✗
✗
✓
✗
✓
43.26
40.36
✗
✗
✗
✓
✓
45.28
42.34
✓
✓
✓
✓
✗
44.24
40.20
Table 3: Ablation of the region selection on UCF-Crime. TIoU of the rectangle selected by the rule of Sec. 3.2 from one map or from their consensus, with or without the temporal smoothing. The grey row is Region-VAD (ours), and the controls are in Table 2 . Bold/underline: best/second-best per column; ties share the mark.
UR-DMU [ 63 ]
MLP [ 41 ]
Tokens
frame
mean
attn.
frame
mean
attn.
backbone
85.94
86.30
86.81
86.35
86.23
86.87
DINOv3
84.31
86.83
85.47
84.77
86.27
84.98
SAM 3
81.99
82.64
82.23
80.62
80.12
79.03
SigLIP 2
86.12
86.79
87.90
85.51
85.98
86.60
Region-VAD (ours)
86.86
87.25
88.93
86.69
86.34
87.01
Table 4: Pooling ablation on UCF-Crime with C-RADIOv4. Frame AUC of each head when its input is the mean of the full-frame tokens (frame), the mean of the tokens inside Rt (mean) or attention over the patches of Rt (attn.), for each projection alone and for the four projections of Region-VAD. Bold/underline: best/second-best per column; ties share the mark.
Figure 4: Per-class localization on UCF-Crime. TIoU of Rt and of the four controls of Table 2 for each anomaly class (number of annotated test videos in parentheses) on the dense boxes of [ 1 ] ; Section B of the appendix repeats it on the boxes of [ 23 ] . Rt exceeds the three box controls in every class but Burglary, whose annotations cover 42% of the frame, and the frame-difference box is ahead of Rt in seven classes and behind in six.
Figure 5: Cumulative IoU with the annotated box. Share of the annotated anomalous test frames whose IoU with the annotated box is below τ (lower is better), for Rt and the four controls on the annotation sets of Table 2 . The value at τ=0 is the share of frames on which the box misses the annotated box entirely. The full-frame box always contains the annotated box, so its IoU is the area share of the annotated box and its curve is the distribution of annotated box sizes. Its value at 0 counts the annotated frames beyond the last segment of a video, which no box reaches.
Figure 6: Qualitative results of Region-VAD on two UCF-Crime test videos. Top: three frames per video with their index and score. Middle: the change field m~t of the same segments, with Rt (solid) and the annotation of [ 1 ] (dashed). Bottom: score of the UR-DMU head, with the annotated span shaded and the three frames marked.
TIoU
UR-DMU [ 63 ]
Backbone
TIoU [ 23 ]
TIoU [ 1 ]
Rt
frame
Region-VAD (ours)
45.71
42.45
88.93
85.94
backbone map
42.75
40.71
86.81
85.94
DINOv3 map
45.60
42.45
85.47
84.31
SAM 3 map
43.26
40.36
82.23
81.99
SigLIP 2 map
45.28
42.34
87.90
86.12
Table 5: Backbone ablation on UCF-Crime. TIoU of the region selected by the rule of Sec. 3.2 from each token set, and frame AUC of UR-DMU with attention pooling inside Rt ( Rt ) or the mean of the full-frame tokens (frame), which for Region-VAD is the full-frame control of Table 2 on the backbone tokens. Indented rows select the region from one map of C-RADIOv4 alone, as in Table 3 , and their detection pools that map’s tokens inside the Rt of Region-VAD, as in Table 4 . The other encoders run the whole pipeline on their single change map. Bold/underline: best/second-best per column; ties share the mark.
Appendix figures & tables4 assets
Supplementary material from the paper’s appendix.
Appendix
UCF-Crime AUC
MSAD AUC
XD-Violence AP
UBnormal AUC
ShanghaiTech AUC
Full frame
85.94 ± 0.15
87.98 ± 0.91
71.44 ± 3.17
63.52 ± 1.27
96.96 ± 0.15
Centred box
84.84 ± 0.50
89.03 ± 0.18
75.03 ± 2.15
59.88 ± 1.43
95.61 ± 0.92
Fixed 8×8 -patch box
85.48 ± 0.32
88.39 ± 0.82
71.70 ± 2.88
64.01 ± 1.71
96.55 ± 0.21
Frame-difference box
85.65 ± 0.22
89.55 ± 0.97
76.83 ± 1.23
67.45 ± 1.41
94.90 ± 0.22
Region-VAD, mean pooling
87.25 ± 0.43
90.80 ± 1.22
79.73 ± 2.36
67.83 ± 1.12
95.92 ± 0.65
Region-VAD (ours)
88.93 ± 1.21
91.39 ± 0.92
81.50 ± 1.41
68.17 ± 3.18
97.13 ± 0.61
Appendix
Table S1 : Spread over seeds. Mean and standard deviation over three training seeds of the detection rows of Table 2 . Bold: best per column.
box area, % of the frame
share of frames
Annotation set
videos
frames
mean
median
IQR
> 50%
< 5%
UCF-Crime, boxes of [ 23 ] (every 16th frame)
140
3,102
23.1
20.1
9.9–33.5
6%
8%
UCF-Crime, dense boxes of [ 1 ]
140
82,584
28.4
23.5
12.1–41.2
21%
3%
MSAD, dense boxes of [ 1 ]
120
31,278
28.0
19.4
7.7–44.4
18%
15%
XD-Violence, dense boxes of [ 1 ]
500
531,398
69.1
76.2
49.1–97.8
74%
3%
UBnormal, official masks
158
68,098
21.8
17.1
8.8–28.7
8%
13%
Appendix
Table S2 : Size of the annotated boxes. Area of the ground-truth box as a percentage of the frame over the annotated anomalous test frames: mean, median, interquartile range, and the share of frames whose box exceeds half the frame or is below 5% of it.
UCF-Crime
Box source
TIoU [ 23 ]
TIoU [ 1 ]
FPS
frame-difference box
45.96
41.25
721
optical flow [ 43 ]
25.14
22.77
61
optical flow, median flow removed
24.40
22.01
61
Region-VAD
45.71
42.45
715
Appendix
Table S3 : Region selection from motion and from tokens on UCF-Crime. TIoU of the rectangle selected by the rule of Sec. 3.2 , with the same temporal smoothing, from pixel frame differencing, from RAFT-large optical-flow magnitude over the 16 frame pairs of each segment, each frame with its predecessor (raw, and with the per-pair median flow removed at a cost negligible next to RAFT), and from the consensus of the four projections (Region-VAD). FPS counts video frames per second of the whole pipeline, encoder and detection head included, on one NVIDIA L4 with segments processed one at a time and video decoding excluded. On an A100, Region-VAD runs at 1,076 FPS (Sec. 4.1 ). Bold/underline: best/second-best per column.
Figure S1 : Per-class localization on UCF-Crime on the boxes of [ 23 ] . Counterpart of Fig. 4 on every 16th frame. TIoU of Rt and of the four controls of Table 2 for each anomaly class (number of annotated test videos in parentheses). Rt exceeds the three box controls in every class, and the frame-difference box is ahead of Rt in eight classes and behind in five.