Grant: Engage Grants Program
McMaster University
$25,000.00
No reference video quality assessment using deep neural networks
FY 2016-17
Summary
The perceptual quality of a video sequence might degrade as a result of compression and/or transmission.Video quality assessment schemes seek to gauge the perceptual quality of a video sequence. In full-referencequality assessment the reference (original) video is available and a direct comparison between the video undertest and the reference is performed. However, in many practical video transmission applications the referencevideo is not available. This necessitates the usage of no-reference video quality assessment. No-referencemeasures can quantify video degradations by exploiting features that are discriminant for degradations.Typically, these features are extracted in transform domains such as wavelet or the DCT. These methods areusually very slow due to the use of computationally expensive transformations and the their outputs do notcorrelate well with subjective assessments. We propose developing a deep learning-based video qualityassessment framework. Recently, deep neural networks have gained researchers' attention and achieved greatsuccess in various vision tasks such as object recognition Our proposed system learns the local patterns andextracts features from sequences of frames in video. With the learned features, we can capture complicatedlocal patterns in videos and exploit task specific information, resulting in more expressive features. This willresult in quality assessments that are more inline with subjective evaluations.
Program
Engage Grants Program
Recipient
Location
Ontario, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data
As published
Application id
505453-2016
Name nom
Shirani, Shahram
Department département
Electrical and Computer Engineering
Organization id
27
Institution établissement
McMaster University
Province en
Ontario
Province fr
Ontario
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2016
Competition year année de concours
2016
Award amount
25000
Program id
EGP
Program name en
Engage Grants Program
Program name fr
Programme de subventions d'engagement partenariat
Group en
Research partnerships programs
Group fr
Programmes de partenariat
Committee code
1553
Committee name en
Ontario Internal Decision Committee
Committee name fr
Comité de décision interne pour l'Ontario
Area of application code
805
Area of application group en
Information and communication services
Area of application group fr
Services d'information et de télécommunications
Area of application en
Communications technologies (satellites, radar, etc.)
Area of application fr
Technologie des télécommunications (satellites, radar,etc.)
Research subject code
2708
Research subject group en
Information technology
Research subject group fr
Technologie de l'information
Research subject en
Image processing
Research subject fr
Traitement des images
Partie
1
Nb partie
1
Application title
No reference video quality assessment using deep neural networks
Keyword
NSERC
Application summary
The perceptual quality of a video sequence might degrade as a result of compression and/or transmission.Video quality assessment schemes seek to gauge the perceptual quality of a video sequence. In full-referencequality assessment the reference (original) video is available and a direct comparison between the video undertest and the reference is performed. However, in many practical video transmission applications the referencevideo is not available. This necessitates the usage of no-reference video quality assessment. No-referencemeasures can quantify video degradations by exploiting features that are discriminant for degradations.Typically, these features are extracted in transform domains such as wavelet or the DCT. These methods areusually very slow due to the use of computationally expensive transformations and the their outputs do notcorrelate well with subjective assessments. We propose developing a deep learning-based video qualityassessment framework. Recently, deep neural networks have gained researchers' attention and achieved greatsuccess in various vision tasks such as object recognition Our proposed system learns the local patterns andextracts features from sequences of frames in video. With the learned features, we can capture complicatedlocal patterns in videos and exploit task specific information, resulting in more expressive features. This willresult in quality assessments that are more inline with subjective evaluations.