5.3 A Data-Compressive 1.5b/2.75b Log-Gradient QVGA Image Sensor with Multi-Scale Readout for Always-On Object Detection. Young, C., Omid-Zohoor, A., Lajevardi, P., & Murmann, B. In 2019 IEEE International Solid- State Circuits Conference - (ISSCC), pages 98–100, February, 2019. doi abstract bibtex Histograms of Oriented Gradients (HOG) are attractive features for object detection in embedded vision applications, as they provide a good trade-off between complexity and detection accuracy. A custom 8b CMOS imager that computes these features on-chip consumes only 52pJ/pixel [1]. However, a complete system also requires a backend detection algorithm, which consumes 940pJ/pixel in an optimized implementation [2]. As shown in the system study of [3], this imbalance mostly stems from the large amount of data seen by the detector. To remedy this issue, the work of [3] studies a feature-extraction approach that aggressively log-quantizes the data, thereby eliminating unnecessary illumination-related bits from the histograms. The custom log-gradient image sensor described in this paper demonstrates this concept in CMOS. It consumes 127pJ/pixel and offers two log-gradient modes (1.5b and 2.75b) along with multi-scale readout. With several algorithmic enhancements enabled by the log gradients as described in [3], the resulting HOG feature compression ratios are 25× (1.5b) and 9.5× (2.75b) when compared to a standard 8b image (without image pyramid). Using 1.5b log gradients, [3] conservatively estimates a 3.3× reduction in backend detection energy for a deformable parts model (DPM) based detector.
@inproceedings{young_5.3_2019,
title = {5.3 {A} {Data}-{Compressive} 1.5b/2.75b {Log}-{Gradient} {QVGA} {Image} {Sensor} with {Multi}-{Scale} {Readout} for {Always}-{On} {Object} {Detection}},
doi = {10.1109/ISSCC.2019.8662436},
abstract = {Histograms of Oriented Gradients (HOG) are attractive features for object detection in embedded vision applications, as they provide a good trade-off between complexity and detection accuracy. A custom 8b CMOS imager that computes these features on-chip consumes only 52pJ/pixel [1]. However, a complete system also requires a backend detection algorithm, which consumes 940pJ/pixel in an optimized implementation [2]. As shown in the system study of [3], this imbalance mostly stems from the large amount of data seen by the detector. To remedy this issue, the work of [3] studies a feature-extraction approach that aggressively log-quantizes the data, thereby eliminating unnecessary illumination-related bits from the histograms. The custom log-gradient image sensor described in this paper demonstrates this concept in CMOS. It consumes 127pJ/pixel and offers two log-gradient modes (1.5b and 2.75b) along with multi-scale readout. With several algorithmic enhancements enabled by the log gradients as described in [3], the resulting HOG feature compression ratios are 25× (1.5b) and 9.5× (2.75b) when compared to a standard 8b image (without image pyramid). Using 1.5b log gradients, [3] conservatively estimates a 3.3× reduction in backend detection energy for a deformable parts model (DPM) based detector.},
booktitle = {2019 {IEEE} {International} {Solid}- {State} {Circuits} {Conference} - ({ISSCC})},
author = {Young, C. and Omid-Zohoor, A. and Lajevardi, P. and Murmann, B.},
month = feb,
year = {2019},
pages = {98--100}
}
Downloads: 0
{"_id":"ScavKo4h6PZgMdDns","bibbaseid":"young-omidzohoor-lajevardi-murmann-53adatacompressive15b275bloggradientqvgaimagesensorwithmultiscalereadoutforalwaysonobjectdetection-2019","authorIDs":[],"author_short":["Young, C.","Omid-Zohoor, A.","Lajevardi, P.","Murmann, B."],"bibdata":{"bibtype":"inproceedings","type":"inproceedings","title":"5.3 A Data-Compressive 1.5b/2.75b Log-Gradient QVGA Image Sensor with Multi-Scale Readout for Always-On Object Detection","doi":"10.1109/ISSCC.2019.8662436","abstract":"Histograms of Oriented Gradients (HOG) are attractive features for object detection in embedded vision applications, as they provide a good trade-off between complexity and detection accuracy. A custom 8b CMOS imager that computes these features on-chip consumes only 52pJ/pixel [1]. However, a complete system also requires a backend detection algorithm, which consumes 940pJ/pixel in an optimized implementation [2]. As shown in the system study of [3], this imbalance mostly stems from the large amount of data seen by the detector. To remedy this issue, the work of [3] studies a feature-extraction approach that aggressively log-quantizes the data, thereby eliminating unnecessary illumination-related bits from the histograms. The custom log-gradient image sensor described in this paper demonstrates this concept in CMOS. It consumes 127pJ/pixel and offers two log-gradient modes (1.5b and 2.75b) along with multi-scale readout. With several algorithmic enhancements enabled by the log gradients as described in [3], the resulting HOG feature compression ratios are 25× (1.5b) and 9.5× (2.75b) when compared to a standard 8b image (without image pyramid). Using 1.5b log gradients, [3] conservatively estimates a 3.3× reduction in backend detection energy for a deformable parts model (DPM) based detector.","booktitle":"2019 IEEE International Solid- State Circuits Conference - (ISSCC)","author":[{"propositions":[],"lastnames":["Young"],"firstnames":["C."],"suffixes":[]},{"propositions":[],"lastnames":["Omid-Zohoor"],"firstnames":["A."],"suffixes":[]},{"propositions":[],"lastnames":["Lajevardi"],"firstnames":["P."],"suffixes":[]},{"propositions":[],"lastnames":["Murmann"],"firstnames":["B."],"suffixes":[]}],"month":"February","year":"2019","pages":"98–100","bibtex":"@inproceedings{young_5.3_2019,\n\ttitle = {5.3 {A} {Data}-{Compressive} 1.5b/2.75b {Log}-{Gradient} {QVGA} {Image} {Sensor} with {Multi}-{Scale} {Readout} for {Always}-{On} {Object} {Detection}},\n\tdoi = {10.1109/ISSCC.2019.8662436},\n\tabstract = {Histograms of Oriented Gradients (HOG) are attractive features for object detection in embedded vision applications, as they provide a good trade-off between complexity and detection accuracy. A custom 8b CMOS imager that computes these features on-chip consumes only 52pJ/pixel [1]. However, a complete system also requires a backend detection algorithm, which consumes 940pJ/pixel in an optimized implementation [2]. As shown in the system study of [3], this imbalance mostly stems from the large amount of data seen by the detector. To remedy this issue, the work of [3] studies a feature-extraction approach that aggressively log-quantizes the data, thereby eliminating unnecessary illumination-related bits from the histograms. The custom log-gradient image sensor described in this paper demonstrates this concept in CMOS. It consumes 127pJ/pixel and offers two log-gradient modes (1.5b and 2.75b) along with multi-scale readout. With several algorithmic enhancements enabled by the log gradients as described in [3], the resulting HOG feature compression ratios are 25× (1.5b) and 9.5× (2.75b) when compared to a standard 8b image (without image pyramid). Using 1.5b log gradients, [3] conservatively estimates a 3.3× reduction in backend detection energy for a deformable parts model (DPM) based detector.},\n\tbooktitle = {2019 {IEEE} {International} {Solid}- {State} {Circuits} {Conference} - ({ISSCC})},\n\tauthor = {Young, C. and Omid-Zohoor, A. and Lajevardi, P. and Murmann, B.},\n\tmonth = feb,\n\tyear = {2019},\n\tpages = {98--100}\n}\n\n","author_short":["Young, C.","Omid-Zohoor, A.","Lajevardi, P.","Murmann, B."],"key":"young_5.3_2019","id":"young_5.3_2019","bibbaseid":"young-omidzohoor-lajevardi-murmann-53adatacompressive15b275bloggradientqvgaimagesensorwithmultiscalereadoutforalwaysonobjectdetection-2019","role":"author","urls":{},"downloads":0},"bibtype":"inproceedings","biburl":"https://bibbase.org/zotero/ky25","creationDate":"2019-05-11T17:47:04.332Z","downloads":0,"keywords":[],"search_terms":["data","compressive","75b","log","gradient","qvga","image","sensor","multi","scale","readout","always","object","detection","young","omid-zohoor","lajevardi","murmann"],"title":"5.3 A Data-Compressive 1.5b/2.75b Log-Gradient QVGA Image Sensor with Multi-Scale Readout for Always-On Object Detection","year":2019,"dataSources":["XxiQtwZYfozhQmvGR"]}