From Wrecks to Wisdom: Recovering Crash Mechanics from Real-World Multi-View Photos

Ondřej Valach; Václav Diviš; Ivan Gruber

cs.CVSep 30, 2026

From Wrecks to Wisdom: Recovering Crash Mechanics from Real-World Multi-View Photos

Authors: Ondřej Valach, Václav Diviš, Ivan Gruber

Organizations: University of West Bohemia, Faculty of Applied Sciences, Department of Cybernetics and New Technologies for the Information Society

Abstract

Estimating accident mechanics from real-world crashes is important for vehicle-safety analysis, injury modeling, crash-severity prediction, and operational workflows such as insurance claim triage. In standard crash records, key metadata such as impact configuration, principal direction of force, and change in velocity ( $ΔV$ ) may be missing, delayed, or corrupted, while post-crash photographs are widely available and contain rich visual evidence of deformation. We study how much crash-mechanics information can be recovered directly from vehicle photos when structured signals are absent. We formulate crash understanding as supervised prediction from per-case multi-view photo sets. Targets include six Collision Deformation Classification (CDC) descriptors and the longitudinal and lateral components of reconstructed $ΔV$ . Each photo is encoded by a shared visual backbone, and the resulting view-level features are fused into a case-level representation from which target-specific heads predict crash descriptors. Using 15.2k training cases from the Crash Investigation Sampling System, drawn from about 1.5M photos before filtering, together with 1.15k validation and 1.15k test cases, we define an evaluation protocol for vision-based crash descriptor estimation from incomplete multi-view evidence. Post-crash imagery alone provides usable signal for several non-trivial crash-mechanics descriptors, while weakly observable and long-tailed targets remain challenging. Within the compared training regimes, the selected joint-training recipe reduces mean absolute angular error for principal direction of force from 20.1 to 14.05 degrees and longitudinal $ΔV$ MAE from 8.04 to 7.45 km/h. Our work provides a reference point for future multimodal fusion with structured crash metadata.

Figures & tables

Figure 1: Comparison of the selected joint-training and single-task reference pipelines. Performance is shown as relative change (%) with respect to the corresponding selected single-task result for each target.

Figure 2: Representative crash cases in the fixed multi-view input format. Each case is shown as an ordered subset of available viewpoints together with its available crash descriptor annotations (CDC-derived targets and $\Delta V$ when available). The descriptor denoted as $\Delta V$ represents the resulting compound vector and is included only for reader interpretation.

Figure 3: Multi-view crash descriptor architecture: a shared SwinV2 encodes nine ordered views, a lightweight fusion transformer aggregates them into a case embedding, and a decoder predicts CDC descriptors and $\Delta V$ .

Target	Out	Classes	Miss%	Top1%	Metrics
Impact plane	Classification	4	0.0	64.9	Acc
DoF ${}_{\text{clock}}$	Circ-Smooth-Cls	12	2.9	46.1	AngErr
Long./lat. zone	Classification	9	0.0	38.3	Acc
Vert./lat. zone	Classification	7	0.0	89.8	Acc
Damage distribution	Classification	8	0.0	75.4	Acc
Deformation extent	Smooth-Cls	9	14.7	43.6	Acc@ $\pm$ 1

Table 1: Target summary with missing-label rates (Miss%) and share of the most frequent class among valid labels (Top1%). Circ-Smooth-Cls: circular classification with Gaussian-smoothed targets and KL divergence [ 10 ] . Smooth-Cls: classification with label smoothing. Acc: accuracy. AngErr: mean absolute angular error (degrees). MAE: mean absolute error. Acc@

\pm

1: accuracy within one neighboring extent bin of the ground-truth label.

Descriptor	Metric	Single-task	Multi-task	$\boldsymbol{\Delta}$ (ST $\rightarrow$ MT)	Rel. $\boldsymbol{\Delta}$
Impact plane	Acc	0.924	0.922	-0.002	-0.2%
Impact plane	$F1_{\mathrm{macro}}$	0.714	0.720	+0.006	+0.8%
	AngErr ( ∘ )	20.10	14.05	-6.05	-30.1%
DoF clock	Acc@ $\pm$ 1	0.859	0.877	+0.018	+2.1%
Zone Long./lat.	Acc	0.582	0.602	+0.020	+3.4%
Zone Long./lat.	$F1_{\mathrm{macro}}$	0.577	0.532	-0.045	-7.8%

Table 2: Comparison between selected single-task reference models and the selected joint-training recipe.

\Delta

denotes the absolute change from the selected single-task model to the selected joint-training model, and Rel.

\Delta

denotes the corresponding relative change with respect to the single-task result. For AngErr and MAE, lower is better. For Acc/

F1

, higher is better. The comparison should be interpreted as an empirical comparison between selected training regimes, not as an isolated ablation of multi-task supervision.

Appendix figures & tables8 assets

Supplementary material from the paper’s appendix.

Appendix

ID	Backbone	Views	Fuse	Target/Loss	AngErr ( ∘ ) $\downarrow$
P1	Eff-B4	Rand $K$ (rep)	Concat.	$\sin/\cos$ MSE	$44.1^{\circ}$
P2	Intern-T	Rand $K$ (rep)	Concat.	Circ-smooth KL	$44.6^{\circ}$
P3	Intern-T	Chan. conc.	–	Circ-smooth KL	$37.4^{\circ}$
P4	SwinV2-B	Rand $K$ (perm)	Mean pool	$\sin/\cos$ MSE	$32.9^{\circ}$
P5	SwinV2-B	Rand $K$ (rep)	Concat.	$\sin/\cos$ MSE	$27.3^{\circ}$
P6	SwinV2-B	Fixed-9 (pad)	Transf.fuse	Circ-smooth KL	$21.8^{\circ}$

Appendix

Table 1: DoF ablations. Eff-B4: EfficientNet-B4. SwinV2-B: SwinV2-base. Intern-T: InternImage-T. Rand

K

: random subset of available views. Rep: replication if fewer than

K

views. Perm: random permutation. Fixed-9: fixed ordered 9-slot input. Pad: zero-padding for missing views and edge padding to preserve image aspect ratio. AngErr: mean absolute angular error (degrees) under circular distance.

Figure 1: Overview of the crash case acquisition and curation pipeline. Starting from publicly available NHTSA CISS cases, we applied data cleaning, metadata-based image filtering, YOLOv8s vehicle filtering, and a dedicated RetinaNet-based wheel-detector filtering stage to obtain the final curated crash-image sets used in this study.

Figure 2: Representative examples removed by the wheel-filtering step. The discarded images are dominated by close-up wheel views and contain limited information about global crash deformation.

Slot construction	Input organization	AngErr ( ∘ ) $\downarrow$
Random / unordered	Random available views	25.18
Metadata slots	Fixed 9-slot, CISS metadata	20.10
Auto-orientation slots	Fixed 9-slot, predicted orientation	20.10

Appendix

Table 2: Effect of view-slot construction on DoF prediction. Metadata slots use CISS viewpoint labels and are used in the main experiments to isolate crash-descriptor learning from viewpoint-estimation noise. Auto-orientation slots use an image-based vehicle-orientation estimator trained from the Car Full View Dataset [ 1 ] to assign views to the fixed 9-slot representation. Results are averaged over three runs using the same DoF model and evaluation protocol.

Figure 3: Label distributions in our CISS-based dataset. We show the class frequencies of all CDC-derived targets together with the distributions of

\Delta V_{\text{long}}

and

\Delta V_{\text{lat}}

(WinSMASH reconstruction). The compound

\Delta V

vector is included for reader interpretation. The categorical targets show strong long-tail behavior with dominant frontal/central configurations.

	Single-task	Multi-task
Epochs	50	50
Learning rate	$1\!\times\!10^{-5}$	$2/6/30\!\times\!10^{-6}$
LR split	–	backbone / fusion+pre-head / task heads
Warmup / anneal	3 / 7	5 / 20
Batch / accum.	4 / –	1 / 4
Effective batch	4	4

Appendix

Table 3: Representative training configurations for the reported single-task and multi-task results. Both regimes use the same case representation, backbone, and fusion module, but differ in initialization, optimization, and loss balancing. Single-task hyperparameters were tuned per target, so that column is a compact summary rather than one identical configuration for all descriptors. Quant.-strat. rare/update: quantile-stratified update-level sampling with at least one rare sample per effective optimizer update. Task-aware WRS: WeightedRandomSampler using the maximum available inverse-frequency weight across categorical targets. Init. strategy: transfer initializes from a related single-task checkpoint, e.g., impact plane for vertical/lateral finetuning; clock-pretrain + random initializes the shared trunk from DoF

{}_{\text{clock}}

pretraining and the remaining components randomly.

Work	Input	Target	Acc
Silver et al. [ 25 ]	1 img	LOC (f/b)	0.920
Ours (ST)	$\leq 9$ images	CDC (f/r/l/b)	0.924
Ours (MT)	$\leq 9$ images	CDC (f/r/l/b)	0.922

Appendix

Table 4: Impact plane image prediction. Silver et al. [ 25 ] evaluate in a curated single-image setting where the background is masked to retain only the vehicle, inputs are limited to front or rear views, and the label space is analogously reduced to front/back location. In contrast, our benchmark uses real-world crash cases represented as multi-view photo sets with natural viewpoint missingness and 4-plane coverage (front/right/left/back).

Work	Input	Scope	Metric	Target	Score [km/h]
Silver et al. [ 25 ]	1 image	R1	MAE	$\Delta V_{\text{long}}$	4.19
Hasija et al. [ 8 ]	1 image, meta	R2	RMSE	$\Delta V_{\text{long}}$	6.38
			MAE	$(\Delta V_{\text{long}},\Delta V_{\text{lat}})$	8.04 / 5.31
Ours (ST)	$\leq 9$ images	R3	RMSE	$(\Delta V_{\text{long}},\Delta V_{\text{lat}})$	11.96 / 8.20
			MAE	$(\Delta V_{\text{long}},\Delta V_{\text{lat}})$	7.45 / 5.09
Ours (MT)	$\leq 9$ images	R3	RMSE	$(\Delta V_{\text{long}},\Delta V_{\text{lat}})$	11.26 / 7.25

Appendix

Table 5:

\Delta V

estimation results. Scope codes: R1 (Silver et al. [ 25 ] ) uses a manually curated set (combined from simulation and real world) of small passenger vehicles for front/rear collisions (

<96

km/h), images are grayscale and manually cropped to the vehicle, reporting MAE for

\Delta V_{\text{long}}

. R2 (Hasija et al. [ 8 ] ) uses a curated frontal-only subset and incorporates vehicle metadata (weight, body type, stiffness); we report their main RMSE for

\Delta V_{\text{long}}

on the full test set (6.38 km/h). They additionally report 7.50 km/h on the EDR-verified subset. R3 (ours) evaluates on real-world CISS multi-view photo sets with natural missingness and image-based inputs, we predict component-wise

(\Delta V_{\text{long}},\Delta V_{\text{lat}})

and report MAE and RMSE, with missing labels handled by masking and a 4-way impact plane.

Explore similar work

SynCrash: A Multi-Stage Pipeline for Zero-Shot Accident Detection and Localization in Traffic Surveillance Video
Aug 30, 2026Arkya Jyoti Bagchi, Ritul Jangir, Varun RaskarAccident Source Term Evaluation Code Video Surveillance
Indian Institute of Technology Jodhpur, India
Zero-Shot Traffic Accident Detection via a Coarse-to-Fine VLM-Tracking Pipeline
Aug 9, 2026Dipit Saha, Shah Mohammad Abdul Mannan, Mohammad Raihan Rashid +2Accident Source Term Evaluation Code Accidents
Bangladesh University of Engineering and Technology
Metadata-Aware Multi-Prompt Reasoning for Zero-Shot Accident Understanding
Jun 10, 2026Tarandeep Singh, Soumyanetra Pal, Soham Biswas +1Accident Source Term Evaluation Code Accidents
Netradyne

Aug 30, 2026, cs.CVJ/K move · Enter open · S save

SynCrash: A Multi-Stage Pipeline for Zero-Shot Accident Detection and Localization in Traffic Surveillance Video

Arkya Jyoti Bagchi, Ritul Jangir, Varun Raskar

Indian Institute of Technology Jodhpur, India

Accident Source Term Evaluation Code Video Surveillance Localization

Open paper AskPDF arXiv

We present SynCrash, a multi-stage pipeline for zero-shot accident detection, spatial localization, and collision-type classification in fixed-view CCTV surveillance video. Our approach addresses the ACCIDENT at CVPR 2026 Challenge, which requires predicting when an accident occurs, where in the frame the impact happens, and what type of collision it is, all without access to labeled real-world training data. The pipeline operates in three decoupled stages: (1) Temporal localization via a VideoMAEv2-giant backbone fine-tuned on CARLA-based synthetic clips with metadata-aware embeddings and dense sliding-window inference; (2) Spatial localization using YOLO for object detection combined with a physics-informed hybrid heuristic that leverages bounding-box overlap and trajectory-based reasoning to predict the impact point; and (3) Collision-type classification using a lightweight rule-based strategy derived from the number and configuration of detected vehicles. The key insight is that temporal understanding benefits from supervised fine-tuning on synthetic data, whereas spatial understanding is better served by pretrained object detectors and physics priors that transfer naturally across domains.

Abstract

Explore similar work

SynCrash: A Multi-Stage Pipeline for Zero-Shot Accident Detection and Localization in Traffic Surveillance Video

Zero-Shot Traffic Accident Detection via a Coarse-to-Fine VLM-Tracking Pipeline

Metadata-Aware Multi-Prompt Reasoning for Zero-Shot Accident Understanding