Towards Real-Time Automatic Portrait Matting on Mobile Devices

Seo, Seokjun; Choi, Sukgeun; Keršner, Martin; Shin, Beomjun; Yoon, Hyung-Suk; Byun, Hyeongmin; Ha, Sungjoo

doi:10.48550/arxiv.1904.03816

Cited by 2 publications

(3 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Seo et al [23] explore an approach targeting real-time execution on mobile devices. Their network is lightweight, using depthwise-separable convolutions and weight quantization.…”

Section: Image Mattingmentioning

confidence: 99%

See 1 more Smart Citation

Temporally Coherent Person Matting Trained on Fake-Motion Dataset

Molodetskikh¹,

Ерофеев²,

Moskalenko³

et al. 2021

Preprint

View full text Add to dashboard Cite

We propose a novel neural-network-based method to perform matting of videos depicting people that does not require additional user input such as trimaps. Our architecture achieves temporal stability of the resulting alpha mattes by using motion-estimation-based smoothing of image-segmentation algorithm outputs, combined with convolutional-LSTM modules on U-Net skip connections.We also propose a fake-motion algorithm that generates training clips for the videomatting network given photos with ground-truth alpha mattes and background videos. We apply random motion to photos and their mattes to simulate movement one would find in real videos and composite the result with the background clips. It lets us train a deep neural network operating on videos in an absence of a large annotated video dataset and provides ground-truth training-clip foreground optical flow for use in loss functions.

show abstract

“…Seo et al [23] explore an approach targeting real-time execution on mobile devices. Their network is lightweight, using depthwise-separable convolutions and weight quantization.…”

Section: Image Mattingmentioning

confidence: 99%

“…COSNet [17], despite being a video method, showed considerable flickering. MMNet [23] exhibited the worst result, likely because it's unsuited to large images. In most comparisons, our proposed approach was preferred over the rest.…”

Section: Subjective Evaluationmentioning

confidence: 99%

Temporally Coherent Person Matting Trained on Fake-Motion Dataset

Molodetskikh¹,

Ерофеев²,

Moskalenko³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…It contains 2000 images of 600 × 800 resolution where 1700 and 300 images are split as training and testing set respectively. To overcome the lack of training data, we augment images by utilizing rotation and left-right flip, as suggested in [36]. Each training image is rotated by [−15 • , 15 • ] in steps of 5 • and left-right flipped, which means that a total of 23800 training images are obtained.…”

Section: Datasetmentioning

confidence: 99%

Attention-guided Image Compression by Deep Reconstruction of Compressive Sensed Saliency Skeleton

Zhang¹,

Wu²

2021

Preprint

View full text Add to dashboard Cite

We propose a deep learning system for attention-guided dual-layer image compression (AGDL). In the AGDL compression system, an image is encoded into two layers, a base layer and an attention-guided refinement layer. Unlike the existing ROI image compression methods that spend an extra bit budget equally on all pixels in ROI, AGDL employs a CNN module to predict those pixels on and near a saliency sketch within ROI that are critical to perceptual quality. Only the critical pixels are further sampled by compressive sensing (CS) to form a very compact refinement layer. Another novel CNN method is developed to jointly decode the two compression layers for a much refined reconstruction, while strictly satisfying the transmitted CS constraints on perceptually critical pixels. Extensive experiments demonstrate that the proposed AGDL system advances the state of the art in perception-aware image compression.

show abstract

Towards Real-Time Automatic Portrait Matting on Mobile Devices

Cited by 2 publications

References 36 publications

Temporally Coherent Person Matting Trained on Fake-Motion Dataset

Temporally Coherent Person Matting Trained on Fake-Motion Dataset

Attention-guided Image Compression by Deep Reconstruction of Compressive Sensed Saliency Skeleton

Contact Info

Product

Resources

About