Subtle Motion Blur Detection and Segmentation from Static Image Artworks

Ganesh Samarth; Sibendu Paul; Solale Tabarestani; Caren Chen

cs.CVFeb 21, 2026

Subtle Motion Blur Detection and Segmentation from Static Image Artworks

Authors: Ganesh Samarth, Sibendu Paul, Solale Tabarestani, Caren Chen

Organizations: Amazon Prime Video

Abstract

Streaming services serve hundreds of millions of viewers worldwide, where visual assets such as thumbnails, box art, and cover images are critical for engagement. Subtle motion blur remains a pervasive quality issue, reducing visual clarity and negatively affecting user trust and click-through rates. However, motion blur detection from static images is underexplored, as existing methods and datasets focus on severe blur and lack fine-grained pixel-level annotations needed for quality-critical applications. Benchmarks such as GOPRO and NFS are dominated by strong synthetic blur and often contain residual blur in their sharp references, leading to ambiguous supervision. We propose SMBlurDetect, a unified framework combining high-quality motion blur specific dataset generation with an end-to-end detector capable of zero-shot detection at multiple granularities. Our pipeline synthesizes realistic motion blur from super high resolution aesthetic images using controllable camera and object motion simulations over SAM segmented regions, enhanced with alpha-aware compositing and balanced sampling to generate subtle, spatially localized blur with precise ground truth masks. We train a U-Net based detector with ImageNet pretrained encoders using a hybrid mask and image centric strategy incorporating curriculum learning, hard negatives, focal loss, blur frequency channels, and resolution aware augmentation.Our method achieves strong zero-shot generalization, reaching 89.68% accuracy on GoPro (vs 66.50% baseline) and 59.77% Mean IoU on CUHK (vs 9.00% baseline), demonstrating 6.6x improvement in segmentation. Qualitative results show accurate localization of subtle blur artifacts, enabling automated filtering of low quality frames and precise region of interest extraction for intelligent cropping.

Figures & tables

Panel of Figure 1 — Figure 1: Examples from GOPRO [ 17 ] and NFS [ 7 ] where “sharp” images still contain subtle motion blur, illustrating why existing datasets are unreliable for training motion-blur detectors for quality-critical applications.

Figure 2: End-to-end SMBlurDetect Pipeline Overview. The system comprises three main components: (1) Dataset Preparation: High-quality images from LAION-5B [ 20 ] are processed through SAM-based segmentation to extract foreground instance masks for critical regions (faces, hands, hair), with hybrid Mask-Centric and Image-Centric sampling strategies controlled by $mask_{ratio}$ parameter. (2) Motion Blur Synthesis: Six physically motivated blur types (straight, curved, zoom with rotation, random-walk, edge-ring, rolling) are applied using exposure-based temporal integration and PSF convolution, with alpha-aware compositing and adaptive edge feathering to generate photorealistic blur with precise ground-truth masks. (3) Dual-Head U-Net Architecture: A ResNet-50 ImageNet-pretrained encoder with decoder skip connections produces binary blur segmentation (Mask Head) and continuous blur-intensity maps (Regression Head), trained using composite loss through three-stage progressive curriculum learning, enabling accurate multi-granularity detection of subtle motion blur in artworks.

Figure 3: Foreground instance masks highlighting regions most susceptible to motion blur, for example face, hair, and hands.

Figure 4: Our photorealistic blur augmentation pipeline implements six distinct blur types, each designed to model a specific motion scenario. For every instance, one blur type is applied at varying strengths, enabling multiple motion patterns to coexist within a single image.

Blur Class Performance:
Metric	Ours	Kim et al. [ 8 ]
Pixel Accuracy (%)	72.69	47.11
Mean IoU (%)	59.77	9.00
Weighted IoU (%)	57.19	23.81
Mean Class Accuracy (%)	72.56	54.28
Precision (%)	75.86	96.37

Table 1: Pixel-level segmentation performance on CUHK dataset [ 21 ]

Metric	Ours	Kim et al. [ 8 ]
Accuracy (%)	89.68	66.50
Precision (%)	92.54	100.00
Recall (%)	86.32	33.00
Specificity (%)	93.05	100.00
F1 Score (%)	89.32	49.62
ROC AUC	0.94	0.79

Table 2: Binary classification performance on GoPro dataset (2.2K sharp and 2.2K blur images).

Metric	Ours	Kim et al. [ 8 ]
Accuracy (%)	80.33	59.33
Precision (%)	84.79	99.30
Recall (%)	73.93	18.80
Specificity (%)	86.73	99.87
F1 Score (%)	79.00	31.61
ROC AUC	0.85	0.70

Table 3: Binary classification performance on NFS dataset (1.5K sharp and 1.5K blur images).

Explore similar work

Aug 27, 2026

SSMB: Self-Supervised Local Feature Detection under Motion Blur

Keypoint detection under motion blur remains a significant challenge, as blur distorts local image structure and degrades the repeatability of feature localization. Existing approaches either rely on computationally expensive deblur-then-detect pipelines that may introduce restoration artifacts, or learn to regress the image positions of handcrafted keypoints extracted on sharp images, which reflects the assumptions of the handcrafted detector rather than what is truly repeatable under blur. We present SSMB, a deblur-free, self-supervised keypoint detector for motion-blurred images that requires neither handcrafted detectors nor external pseudo-labels. SSMB introduces the Local Discriminability Enhancement (LDE) module, which restores fine-grained local discriminability after global feature mixing. Training is performed in two stages. First, geometric pretraining on synthetic shapes bootstraps spatially discriminative keypoint detection without any external detector, just from the rendered geometry. Second, blur-aware training on real sharp-blur image pairs learns blur-invariant detection through a multi-component self-supervised objective that enforces cross-domain consistency, geometric alignment, and spatial coverage. Extensive evaluations on keypoint detection, image matching, relative pose estimation, and visual localization under motion blur demonstrate that SSMB establishes a new state-of-the-art among sparse keypoint detectors, consistently outperforming both supervised and self-supervised baselines across all tasks. Code, models, and datasets will be publicly available upon paper acceptance.

Zhenjun Zhao, Fabio Bellavia, Wenting Wang +6

University of Zaragoza, Zaragoza, Spain · University of Palermo, Palermo, Italy · The Chinese University of Hong Kong, Hong Kong, China +5

Keypoint Detection Image Matching

Ask PDF

Jul 11, 2026

BOCCHI: A More Realistic and Challenging Benchmark for Local Motion Blur Detection with MSDCT-UNet

Local motion blur detection requires pixel-level localization of blurred regions. Existing benchmarks let models rely on gradient shortcuts that fail to transfer. We introduce BOCCHI (Blurred Objects Captured across Cameras with Human-annotated Imagery), a real-captured benchmark whose sharp regions overlap the blur gradient distribution and defeat these shortcuts, and propose MSDCT-UNet (Multi-Scale Discrete Cosine Transform UNet), a frequency-aware encoder-decoder injecting multi-scale DCT priors through DCT Attention and FiLM. MSDCT-UNet ranks first in in-domain mIoU and boundary localization on BOCCHI, and BOCCHI-trained models outperform every other training source on cross-dataset transfer with only 633 training images. Our project page is available at https://brianchen1120.github.io/project/bocchi/.

Kuan-Lin Chen, Yuan-Kang Lee, Cheng-Yuan Chiang +1

Graduate Institute of Communication Engineering, National Taiwan University, Taipei, Taiwan · MediaTek Inc., Hsinchu, Taiwan

U-Net

Ask PDF

Jun 24, 2025

SloMoDeblur: A Large-Scale Smartphone Image Deblurring Dataset

Motion blur remains one of the most common and visually disruptive degradations in real-world smartphone imaging, yet existing deblurring benchmarks are often limited in scale, resolution, or domain relevance. This gap is especially pronounced for smartphones, where rolling shutter, small sensors, and ISP processing produce blur statistics that differ from GoPro/DSLR-based benchmarks. We introduce a large-scale smartphone-oriented deblurring dataset constructed from 240~fps slow-motion video. To approximate exposure-time radiance integration, we synthesize blur by temporally averaging a fixed window of

N=30

consecutive frames, which corresponds to an effective exposure of

T=1/8

~second, and we select the temporally centered frame as the sharp ground truth. The resulting benchmark contains 42,045 paired blur--sharp images at

1920\times1080

resolution spanning 843 distinct scenes, with a train/test split of 37,841/4,204 pairs. We benchmark multiple state-of-the-art deblurring models using PSNR and SSIM and observe consistent performance degradation relative to the baseline similarity between the input blurry images and ground truth, underscoring the realism and difficulty of the proposed data. We release the dataset and generation scripts via HuggingFace to facilitate the development and evaluation of robust, deployment-oriented deblurring methods.

Syed Mumtahin Mahmud, Mahdi Mohd Hossain Noki, Prothito Shovon Majumder +4

Cognitive Agents and Interaction Lab, Department of Computer Science and Engineering, University of Dhaka, Dhaka-1000, Bangladesh

Image Deblurring

Ask PDF