User:Ahamsala: Difference between revisions
No edit summary |
|||
| Line 1: | Line 1: | ||
One of the major parameter that can be changed in JPEG pipeline is the quantization table, which is the main source of artifacts to be added in the image to make it lossless compression as shown in \cite{li2019adacompress, 10.1007/11552499_16}. The authors got motivated to change in the JPEG configuration to optimize uploading rate of different cloud computer vision without considering pre-knowledge of the original model and dataset. In contrast to the authors in \cite{gueguen2018faster,liu2018deepn, torfason2018image} which they adjust the JPEG configuration according to retrain the parameters or the structure of the model. They considered the lackness of undefining the quantization level which decrease the image rate and quality but the the deep learning model can still recognize it as shown in \cite{10.1007/11552499_16}. The authors in \cite{li2019adacompress} used Deep Reinforcement learning (DRL) in an online manner to choose the quantization level to upload an image to the cloud for computer vision model and this is the only approach to design an adaptive JPEG based on RL mechanism. | == Presented by == | ||
Ahmed Salamah | |||
== Introduction == | |||
One of the major parameter that can be changed in JPEG pipeline is the quantization table, which is the main source of artifacts to be added in the image to make it lossless compression as shown in \cite{li2019adacompress, | |||
10.1007/11552499_16}. The authors got motivated to change in the JPEG configuration to optimize uploading rate of different cloud computer vision without considering pre-knowledge of the original model and dataset. In contrast to the authors in \cite{gueguen2018faster,liu2018deepn, torfason2018image} which they adjust the JPEG configuration according to retrain the parameters or the structure of the model. They considered the lackness of undefining the quantization level which decrease the image rate and quality but the the deep learning model can still recognize it as shown in \cite{10.1007/11552499_16}. The authors in \cite{li2019adacompress} used Deep Reinforcement learning (DRL) in an online manner to choose the quantization level to upload an image to the cloud for computer vision model and this is the only approach to design an adaptive JPEG based on RL mechanism. | |||
Revision as of 03:09, 13 November 2020
Presented by
Ahmed Salamah
Introduction
One of the major parameter that can be changed in JPEG pipeline is the quantization table, which is the main source of artifacts to be added in the image to make it lossless compression as shown in \cite{li2019adacompress, 10.1007/11552499_16}. The authors got motivated to change in the JPEG configuration to optimize uploading rate of different cloud computer vision without considering pre-knowledge of the original model and dataset. In contrast to the authors in \cite{gueguen2018faster,liu2018deepn, torfason2018image} which they adjust the JPEG configuration according to retrain the parameters or the structure of the model. They considered the lackness of undefining the quantization level which decrease the image rate and quality but the the deep learning model can still recognize it as shown in \cite{10.1007/11552499_16}. The authors in \cite{li2019adacompress} used Deep Reinforcement learning (DRL) in an online manner to choose the quantization level to upload an image to the cloud for computer vision model and this is the only approach to design an adaptive JPEG based on RL mechanism.