Two-Stream Action Recognition-Oriented Video Super-Resolution

Zhang, Haochen; Liu, Dong; Xiong, Zhiwei

doi:10.1109/iccv.2019.00889

Cited by 51 publications

(21 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Considering that this is not flexible for HAR, Wang [109] introduced an end-to-end framework that is capable of taking videos of arbitrary sizes and lengths as inputs. To address the problem of HAR using videos at low spatial resolutions, Zhang et al [118] studied the super-resolution problem of video for HAR.…”

Section: Deep Learning Methodsmentioning

confidence: 99%

Human Action Recognition from Various Data Modalities: A Review

Rahmani¹,

Bennamoun²,

Ke³

2021

Preprint

View full text Add to dashboard Cite

show abstract

Section: Deep Learning Methodsmentioning

confidence: 99%

Human Action Recognition from Various Data Modalities: A Review

Rahmani¹,

Bennamoun²,

Ke³

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…In video deblurring, [26] estimated the segmentation probability map first and finished deblurring via the segmentation prior. To improve classification, [28] enhanced the target image and [37] increased the resolution of the input image, and [32] restored the degraded images to clean images in the feature domain of a classifier. These works above focused on the performance of only one task.…”

Section: Synergy Between High-level and Low-level Vision Tasksmentioning

confidence: 99%

Synergy Between Semantic Segmentation and Image Denoising via Alternate Boosting

Xu¹,

Sun²,

Liu³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

The capability of image semantic segmentation may be deteriorated due to noisy input image, where image denoising prior to segmentation helps. Both image denoising and semantic segmentation have been developed significantly with the advance of deep learning. Thus, we are interested in the synergy between them by using a holistic deep model. We observe that not only denoising helps combat the drop of segmentation accuracy due to noise, but also pixel-wise semantic information boosts the capability of denoising. We then propose a boosting network to perform denoising and segmentation alternately. The proposed network is composed of multiple segmentation and denoising blocks (SDBs), each of which estimates semantic map then uses the map to regularize denoising. Experimental results show that the denoised image quality is improved substantially and the segmentation accuracy is improved to close to that of clean images. Our code and models will be made publicly available.

show abstract

“…Generally, most of two-stream type works ( [36], [37], [38]) learn to fuse two outputs of this network. Indeed, we can directly add the connection between outputs of IENet and RENet.…”

Section: B Context-sensitive Decomposition Connectionmentioning

confidence: 99%

Learning Deep Context-Sensitive Decomposition for Low-Light Image Enhancement

Ma¹,

Liu²,

Zhang³

et al. 2021

Preprint

View full text Add to dashboard Cite

Enhancing the quality of low-light images plays a very important role in many image processing and multimedia applications. In recent years, a variety of deep learning techniques have been developed to address this challenging task. A typical framework is to simultaneously estimate the illumination and reflectance, but they disregard the scene-level contextual information encapsulated in feature spaces, causing many unfavorable outcomes, e.g., details loss, color unsaturation, artifacts, and so on. To address these issues, we develop a new context-sensitive decomposition network architecture to exploit the scene-level contextual dependencies on spatial scales. More concretely, we build a two-stream estimation mechanism including reflectance and illumination estimation network. We design a novel context-sensitive decomposition connection to bridge the two-stream mechanism by incorporating the physical principle. The spatially-varying illumination guidance is further constructed for achieving the edge-aware smoothness property of the illumination component. According to different training patterns, we construct CSDNet (paired supervision) and CS-DGAN (unpaired supervision) to fully evaluate our designed architecture. We test our method on seven testing benchmarks (including MIT-Adobe FiveK, LOL, ExDark, NPE, etc.) to conduct plenty of analytical and evaluated experiments. Thanks to our designed context-sensitive decomposition connection, we successfully realized excellent enhanced results (with sufficient details, vivid colors, and few noises), which fully indicates our superiority against existing state-of-the-art approaches. Finally, considering the practical needs for high-efficiency, we develop a lightweight CSDNet (named LiteCSDNet) by reducing the number of channels. Further, by sharing an encoder for these two components, we obtain a more lightweight version (SLiteCSDNet for short). SLiteCSDNet just contains 0.0301M parameters but achieves the almost same performance as CSDNet. Code is available at https://github.com/KarelZhang/CSDNet-CSDGAN.

show abstract

Two-Stream Action Recognition-Oriented Video Super-Resolution

Cited by 51 publications

References 45 publications

Human Action Recognition from Various Data Modalities: A Review

Human Action Recognition from Various Data Modalities: A Review

Synergy Between Semantic Segmentation and Image Denoising via Alternate Boosting

Learning Deep Context-Sensitive Decomposition for Low-Light Image Enhancement

Contact Info

Product

Resources

About