Single shot object detection with refined feature
Abstract Object classification and localization are two significant aspects of object detector based on the Single Shot MultiBox Detector (SSD). In general, the more feature maps there are, the better the object classification performance will be. However, when the information of excessive feature m...
Ausführliche Beschreibung
Autor*in: |
Zhang, Xiaojuan [verfasserIn] |
---|
Format: |
Artikel |
---|---|
Sprache: |
Englisch |
Erschienen: |
2020 |
---|
Schlagwörter: |
---|
Anmerkung: |
© Springer Science+Business Media, LLC, part of Springer Nature 2020 |
---|
Übergeordnetes Werk: |
Enthalten in: Multimedia tools and applications - Springer US, 1995, 80(2020), 1 vom: 04. Sept., Seite 737-752 |
---|---|
Übergeordnetes Werk: |
volume:80 ; year:2020 ; number:1 ; day:04 ; month:09 ; pages:737-752 |
Links: |
---|
DOI / URN: |
10.1007/s11042-020-09483-4 |
---|
Katalog-ID: |
OLC2122573473 |
---|
LEADER | 01000naa a22002652 4500 | ||
---|---|---|---|
001 | OLC2122573473 | ||
003 | DE-627 | ||
005 | 20230505063909.0 | ||
007 | tu | ||
008 | 230505s2020 xx ||||| 00| ||eng c | ||
024 | 7 | |a 10.1007/s11042-020-09483-4 |2 doi | |
035 | |a (DE-627)OLC2122573473 | ||
035 | |a (DE-He213)s11042-020-09483-4-p | ||
040 | |a DE-627 |b ger |c DE-627 |e rakwb | ||
041 | |a eng | ||
082 | 0 | 4 | |a 070 |a 004 |q VZ |
100 | 1 | |a Zhang, Xiaojuan |e verfasserin |0 (orcid)0000-0002-2194-3257 |4 aut | |
245 | 1 | 0 | |a Single shot object detection with refined feature |
264 | 1 | |c 2020 | |
336 | |a Text |b txt |2 rdacontent | ||
337 | |a ohne Hilfsmittel zu benutzen |b n |2 rdamedia | ||
338 | |a Band |b nc |2 rdacarrier | ||
500 | |a © Springer Science+Business Media, LLC, part of Springer Nature 2020 | ||
520 | |a Abstract Object classification and localization are two significant aspects of object detector based on the Single Shot MultiBox Detector (SSD). In general, the more feature maps there are, the better the object classification performance will be. However, when the information of excessive feature maps are sparse and unnecessary, the performance of object detection is slightly improved or maybe precisely opposite, which is instead harmful to the production of object localization. The performance of object detectors is not only related to the number of feature maps but also relies partly on the bounding box regression and Non-Maximum Suppression (NMS). In this paper, a detector is constructed based on SSD, called Detection with Refined Feature (DRF), involving center map and scale map, the detection loss is reshaped. Our motivation is to improve the accuracy of classification and localization by searching for central points and predicting the scales of the object points. Center map is used to predict the Intersection over Union (IoU) between the prediction box and ground truth box, while scale map considers the relationships among the different scales. Experimental results on both Pascal VOC and MS COCO 2014 instance datasets demonstrate the effectiveness of DRF. Using Darknet53, we achieve an 86.4% mean Average Precision (mAP) on Pascal VOC2007 and an 87.4% mAP on Pascal VOC2007 and VOC2012. On MS COCO, the DRF with ResNet50 still achieves moderate improvement. | ||
650 | 4 | |a Object detection | |
650 | 4 | |a SSD | |
650 | 4 | |a Bounding box | |
650 | 4 | |a Center map | |
650 | 4 | |a Scale map | |
700 | 1 | |a Wang, Changying |4 aut | |
700 | 1 | |a Cheng, Li |4 aut | |
700 | 1 | |a Jiang, Shuihan |4 aut | |
700 | 1 | |a Qi, Junting |4 aut | |
773 | 0 | 8 | |i Enthalten in |t Multimedia tools and applications |d Springer US, 1995 |g 80(2020), 1 vom: 04. Sept., Seite 737-752 |w (DE-627)189064145 |w (DE-600)1287642-2 |w (DE-576)052842126 |x 1380-7501 |7 nnns |
773 | 1 | 8 | |g volume:80 |g year:2020 |g number:1 |g day:04 |g month:09 |g pages:737-752 |
856 | 4 | 1 | |u https://doi.org/10.1007/s11042-020-09483-4 |z lizenzpflichtig |3 Volltext |
912 | |a GBV_USEFLAG_A | ||
912 | |a SYSFLAG_A | ||
912 | |a GBV_OLC | ||
912 | |a SSG-OLC-MAT | ||
912 | |a SSG-OLC-BUB | ||
912 | |a SSG-OLC-MKW | ||
951 | |a AR | ||
952 | |d 80 |j 2020 |e 1 |b 04 |c 09 |h 737-752 |
author_variant |
x z xz c w cw l c lc s j sj j q jq |
---|---|
matchkey_str |
article:13807501:2020----::igehtbeteetowtr |
hierarchy_sort_str |
2020 |
publishDate |
2020 |
allfields |
10.1007/s11042-020-09483-4 doi (DE-627)OLC2122573473 (DE-He213)s11042-020-09483-4-p DE-627 ger DE-627 rakwb eng 070 004 VZ Zhang, Xiaojuan verfasserin (orcid)0000-0002-2194-3257 aut Single shot object detection with refined feature 2020 Text txt rdacontent ohne Hilfsmittel zu benutzen n rdamedia Band nc rdacarrier © Springer Science+Business Media, LLC, part of Springer Nature 2020 Abstract Object classification and localization are two significant aspects of object detector based on the Single Shot MultiBox Detector (SSD). In general, the more feature maps there are, the better the object classification performance will be. However, when the information of excessive feature maps are sparse and unnecessary, the performance of object detection is slightly improved or maybe precisely opposite, which is instead harmful to the production of object localization. The performance of object detectors is not only related to the number of feature maps but also relies partly on the bounding box regression and Non-Maximum Suppression (NMS). In this paper, a detector is constructed based on SSD, called Detection with Refined Feature (DRF), involving center map and scale map, the detection loss is reshaped. Our motivation is to improve the accuracy of classification and localization by searching for central points and predicting the scales of the object points. Center map is used to predict the Intersection over Union (IoU) between the prediction box and ground truth box, while scale map considers the relationships among the different scales. Experimental results on both Pascal VOC and MS COCO 2014 instance datasets demonstrate the effectiveness of DRF. Using Darknet53, we achieve an 86.4% mean Average Precision (mAP) on Pascal VOC2007 and an 87.4% mAP on Pascal VOC2007 and VOC2012. On MS COCO, the DRF with ResNet50 still achieves moderate improvement. Object detection SSD Bounding box Center map Scale map Wang, Changying aut Cheng, Li aut Jiang, Shuihan aut Qi, Junting aut Enthalten in Multimedia tools and applications Springer US, 1995 80(2020), 1 vom: 04. Sept., Seite 737-752 (DE-627)189064145 (DE-600)1287642-2 (DE-576)052842126 1380-7501 nnns volume:80 year:2020 number:1 day:04 month:09 pages:737-752 https://doi.org/10.1007/s11042-020-09483-4 lizenzpflichtig Volltext GBV_USEFLAG_A SYSFLAG_A GBV_OLC SSG-OLC-MAT SSG-OLC-BUB SSG-OLC-MKW AR 80 2020 1 04 09 737-752 |
spelling |
10.1007/s11042-020-09483-4 doi (DE-627)OLC2122573473 (DE-He213)s11042-020-09483-4-p DE-627 ger DE-627 rakwb eng 070 004 VZ Zhang, Xiaojuan verfasserin (orcid)0000-0002-2194-3257 aut Single shot object detection with refined feature 2020 Text txt rdacontent ohne Hilfsmittel zu benutzen n rdamedia Band nc rdacarrier © Springer Science+Business Media, LLC, part of Springer Nature 2020 Abstract Object classification and localization are two significant aspects of object detector based on the Single Shot MultiBox Detector (SSD). In general, the more feature maps there are, the better the object classification performance will be. However, when the information of excessive feature maps are sparse and unnecessary, the performance of object detection is slightly improved or maybe precisely opposite, which is instead harmful to the production of object localization. The performance of object detectors is not only related to the number of feature maps but also relies partly on the bounding box regression and Non-Maximum Suppression (NMS). In this paper, a detector is constructed based on SSD, called Detection with Refined Feature (DRF), involving center map and scale map, the detection loss is reshaped. Our motivation is to improve the accuracy of classification and localization by searching for central points and predicting the scales of the object points. Center map is used to predict the Intersection over Union (IoU) between the prediction box and ground truth box, while scale map considers the relationships among the different scales. Experimental results on both Pascal VOC and MS COCO 2014 instance datasets demonstrate the effectiveness of DRF. Using Darknet53, we achieve an 86.4% mean Average Precision (mAP) on Pascal VOC2007 and an 87.4% mAP on Pascal VOC2007 and VOC2012. On MS COCO, the DRF with ResNet50 still achieves moderate improvement. Object detection SSD Bounding box Center map Scale map Wang, Changying aut Cheng, Li aut Jiang, Shuihan aut Qi, Junting aut Enthalten in Multimedia tools and applications Springer US, 1995 80(2020), 1 vom: 04. Sept., Seite 737-752 (DE-627)189064145 (DE-600)1287642-2 (DE-576)052842126 1380-7501 nnns volume:80 year:2020 number:1 day:04 month:09 pages:737-752 https://doi.org/10.1007/s11042-020-09483-4 lizenzpflichtig Volltext GBV_USEFLAG_A SYSFLAG_A GBV_OLC SSG-OLC-MAT SSG-OLC-BUB SSG-OLC-MKW AR 80 2020 1 04 09 737-752 |
allfields_unstemmed |
10.1007/s11042-020-09483-4 doi (DE-627)OLC2122573473 (DE-He213)s11042-020-09483-4-p DE-627 ger DE-627 rakwb eng 070 004 VZ Zhang, Xiaojuan verfasserin (orcid)0000-0002-2194-3257 aut Single shot object detection with refined feature 2020 Text txt rdacontent ohne Hilfsmittel zu benutzen n rdamedia Band nc rdacarrier © Springer Science+Business Media, LLC, part of Springer Nature 2020 Abstract Object classification and localization are two significant aspects of object detector based on the Single Shot MultiBox Detector (SSD). In general, the more feature maps there are, the better the object classification performance will be. However, when the information of excessive feature maps are sparse and unnecessary, the performance of object detection is slightly improved or maybe precisely opposite, which is instead harmful to the production of object localization. The performance of object detectors is not only related to the number of feature maps but also relies partly on the bounding box regression and Non-Maximum Suppression (NMS). In this paper, a detector is constructed based on SSD, called Detection with Refined Feature (DRF), involving center map and scale map, the detection loss is reshaped. Our motivation is to improve the accuracy of classification and localization by searching for central points and predicting the scales of the object points. Center map is used to predict the Intersection over Union (IoU) between the prediction box and ground truth box, while scale map considers the relationships among the different scales. Experimental results on both Pascal VOC and MS COCO 2014 instance datasets demonstrate the effectiveness of DRF. Using Darknet53, we achieve an 86.4% mean Average Precision (mAP) on Pascal VOC2007 and an 87.4% mAP on Pascal VOC2007 and VOC2012. On MS COCO, the DRF with ResNet50 still achieves moderate improvement. Object detection SSD Bounding box Center map Scale map Wang, Changying aut Cheng, Li aut Jiang, Shuihan aut Qi, Junting aut Enthalten in Multimedia tools and applications Springer US, 1995 80(2020), 1 vom: 04. Sept., Seite 737-752 (DE-627)189064145 (DE-600)1287642-2 (DE-576)052842126 1380-7501 nnns volume:80 year:2020 number:1 day:04 month:09 pages:737-752 https://doi.org/10.1007/s11042-020-09483-4 lizenzpflichtig Volltext GBV_USEFLAG_A SYSFLAG_A GBV_OLC SSG-OLC-MAT SSG-OLC-BUB SSG-OLC-MKW AR 80 2020 1 04 09 737-752 |
allfieldsGer |
10.1007/s11042-020-09483-4 doi (DE-627)OLC2122573473 (DE-He213)s11042-020-09483-4-p DE-627 ger DE-627 rakwb eng 070 004 VZ Zhang, Xiaojuan verfasserin (orcid)0000-0002-2194-3257 aut Single shot object detection with refined feature 2020 Text txt rdacontent ohne Hilfsmittel zu benutzen n rdamedia Band nc rdacarrier © Springer Science+Business Media, LLC, part of Springer Nature 2020 Abstract Object classification and localization are two significant aspects of object detector based on the Single Shot MultiBox Detector (SSD). In general, the more feature maps there are, the better the object classification performance will be. However, when the information of excessive feature maps are sparse and unnecessary, the performance of object detection is slightly improved or maybe precisely opposite, which is instead harmful to the production of object localization. The performance of object detectors is not only related to the number of feature maps but also relies partly on the bounding box regression and Non-Maximum Suppression (NMS). In this paper, a detector is constructed based on SSD, called Detection with Refined Feature (DRF), involving center map and scale map, the detection loss is reshaped. Our motivation is to improve the accuracy of classification and localization by searching for central points and predicting the scales of the object points. Center map is used to predict the Intersection over Union (IoU) between the prediction box and ground truth box, while scale map considers the relationships among the different scales. Experimental results on both Pascal VOC and MS COCO 2014 instance datasets demonstrate the effectiveness of DRF. Using Darknet53, we achieve an 86.4% mean Average Precision (mAP) on Pascal VOC2007 and an 87.4% mAP on Pascal VOC2007 and VOC2012. On MS COCO, the DRF with ResNet50 still achieves moderate improvement. Object detection SSD Bounding box Center map Scale map Wang, Changying aut Cheng, Li aut Jiang, Shuihan aut Qi, Junting aut Enthalten in Multimedia tools and applications Springer US, 1995 80(2020), 1 vom: 04. Sept., Seite 737-752 (DE-627)189064145 (DE-600)1287642-2 (DE-576)052842126 1380-7501 nnns volume:80 year:2020 number:1 day:04 month:09 pages:737-752 https://doi.org/10.1007/s11042-020-09483-4 lizenzpflichtig Volltext GBV_USEFLAG_A SYSFLAG_A GBV_OLC SSG-OLC-MAT SSG-OLC-BUB SSG-OLC-MKW AR 80 2020 1 04 09 737-752 |
allfieldsSound |
10.1007/s11042-020-09483-4 doi (DE-627)OLC2122573473 (DE-He213)s11042-020-09483-4-p DE-627 ger DE-627 rakwb eng 070 004 VZ Zhang, Xiaojuan verfasserin (orcid)0000-0002-2194-3257 aut Single shot object detection with refined feature 2020 Text txt rdacontent ohne Hilfsmittel zu benutzen n rdamedia Band nc rdacarrier © Springer Science+Business Media, LLC, part of Springer Nature 2020 Abstract Object classification and localization are two significant aspects of object detector based on the Single Shot MultiBox Detector (SSD). In general, the more feature maps there are, the better the object classification performance will be. However, when the information of excessive feature maps are sparse and unnecessary, the performance of object detection is slightly improved or maybe precisely opposite, which is instead harmful to the production of object localization. The performance of object detectors is not only related to the number of feature maps but also relies partly on the bounding box regression and Non-Maximum Suppression (NMS). In this paper, a detector is constructed based on SSD, called Detection with Refined Feature (DRF), involving center map and scale map, the detection loss is reshaped. Our motivation is to improve the accuracy of classification and localization by searching for central points and predicting the scales of the object points. Center map is used to predict the Intersection over Union (IoU) between the prediction box and ground truth box, while scale map considers the relationships among the different scales. Experimental results on both Pascal VOC and MS COCO 2014 instance datasets demonstrate the effectiveness of DRF. Using Darknet53, we achieve an 86.4% mean Average Precision (mAP) on Pascal VOC2007 and an 87.4% mAP on Pascal VOC2007 and VOC2012. On MS COCO, the DRF with ResNet50 still achieves moderate improvement. Object detection SSD Bounding box Center map Scale map Wang, Changying aut Cheng, Li aut Jiang, Shuihan aut Qi, Junting aut Enthalten in Multimedia tools and applications Springer US, 1995 80(2020), 1 vom: 04. Sept., Seite 737-752 (DE-627)189064145 (DE-600)1287642-2 (DE-576)052842126 1380-7501 nnns volume:80 year:2020 number:1 day:04 month:09 pages:737-752 https://doi.org/10.1007/s11042-020-09483-4 lizenzpflichtig Volltext GBV_USEFLAG_A SYSFLAG_A GBV_OLC SSG-OLC-MAT SSG-OLC-BUB SSG-OLC-MKW AR 80 2020 1 04 09 737-752 |
language |
English |
source |
Enthalten in Multimedia tools and applications 80(2020), 1 vom: 04. Sept., Seite 737-752 volume:80 year:2020 number:1 day:04 month:09 pages:737-752 |
sourceStr |
Enthalten in Multimedia tools and applications 80(2020), 1 vom: 04. Sept., Seite 737-752 volume:80 year:2020 number:1 day:04 month:09 pages:737-752 |
format_phy_str_mv |
Article |
institution |
findex.gbv.de |
topic_facet |
Object detection SSD Bounding box Center map Scale map |
dewey-raw |
070 |
isfreeaccess_bool |
false |
container_title |
Multimedia tools and applications |
authorswithroles_txt_mv |
Zhang, Xiaojuan @@aut@@ Wang, Changying @@aut@@ Cheng, Li @@aut@@ Jiang, Shuihan @@aut@@ Qi, Junting @@aut@@ |
publishDateDaySort_date |
2020-09-04T00:00:00Z |
hierarchy_top_id |
189064145 |
dewey-sort |
270 |
id |
OLC2122573473 |
language_de |
englisch |
fullrecord |
<?xml version="1.0" encoding="UTF-8"?><collection xmlns="http://www.loc.gov/MARC21/slim"><record><leader>01000naa a22002652 4500</leader><controlfield tag="001">OLC2122573473</controlfield><controlfield tag="003">DE-627</controlfield><controlfield tag="005">20230505063909.0</controlfield><controlfield tag="007">tu</controlfield><controlfield tag="008">230505s2020 xx ||||| 00| ||eng c</controlfield><datafield tag="024" ind1="7" ind2=" "><subfield code="a">10.1007/s11042-020-09483-4</subfield><subfield code="2">doi</subfield></datafield><datafield tag="035" ind1=" " ind2=" "><subfield code="a">(DE-627)OLC2122573473</subfield></datafield><datafield tag="035" ind1=" " ind2=" "><subfield code="a">(DE-He213)s11042-020-09483-4-p</subfield></datafield><datafield tag="040" ind1=" " ind2=" "><subfield code="a">DE-627</subfield><subfield code="b">ger</subfield><subfield code="c">DE-627</subfield><subfield code="e">rakwb</subfield></datafield><datafield tag="041" ind1=" " ind2=" "><subfield code="a">eng</subfield></datafield><datafield tag="082" ind1="0" ind2="4"><subfield code="a">070</subfield><subfield code="a">004</subfield><subfield code="q">VZ</subfield></datafield><datafield tag="100" ind1="1" ind2=" "><subfield code="a">Zhang, Xiaojuan</subfield><subfield code="e">verfasserin</subfield><subfield code="0">(orcid)0000-0002-2194-3257</subfield><subfield code="4">aut</subfield></datafield><datafield tag="245" ind1="1" ind2="0"><subfield code="a">Single shot object detection with refined feature</subfield></datafield><datafield tag="264" ind1=" " ind2="1"><subfield code="c">2020</subfield></datafield><datafield tag="336" ind1=" " ind2=" "><subfield code="a">Text</subfield><subfield code="b">txt</subfield><subfield code="2">rdacontent</subfield></datafield><datafield tag="337" ind1=" " ind2=" "><subfield code="a">ohne Hilfsmittel zu benutzen</subfield><subfield code="b">n</subfield><subfield code="2">rdamedia</subfield></datafield><datafield tag="338" ind1=" " ind2=" "><subfield code="a">Band</subfield><subfield code="b">nc</subfield><subfield code="2">rdacarrier</subfield></datafield><datafield tag="500" ind1=" " ind2=" "><subfield code="a">© Springer Science+Business Media, LLC, part of Springer Nature 2020</subfield></datafield><datafield tag="520" ind1=" " ind2=" "><subfield code="a">Abstract Object classification and localization are two significant aspects of object detector based on the Single Shot MultiBox Detector (SSD). In general, the more feature maps there are, the better the object classification performance will be. However, when the information of excessive feature maps are sparse and unnecessary, the performance of object detection is slightly improved or maybe precisely opposite, which is instead harmful to the production of object localization. The performance of object detectors is not only related to the number of feature maps but also relies partly on the bounding box regression and Non-Maximum Suppression (NMS). In this paper, a detector is constructed based on SSD, called Detection with Refined Feature (DRF), involving center map and scale map, the detection loss is reshaped. Our motivation is to improve the accuracy of classification and localization by searching for central points and predicting the scales of the object points. Center map is used to predict the Intersection over Union (IoU) between the prediction box and ground truth box, while scale map considers the relationships among the different scales. Experimental results on both Pascal VOC and MS COCO 2014 instance datasets demonstrate the effectiveness of DRF. Using Darknet53, we achieve an 86.4% mean Average Precision (mAP) on Pascal VOC2007 and an 87.4% mAP on Pascal VOC2007 and VOC2012. On MS COCO, the DRF with ResNet50 still achieves moderate improvement.</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Object detection</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">SSD</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Bounding box</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Center map</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Scale map</subfield></datafield><datafield tag="700" ind1="1" ind2=" "><subfield code="a">Wang, Changying</subfield><subfield code="4">aut</subfield></datafield><datafield tag="700" ind1="1" ind2=" "><subfield code="a">Cheng, Li</subfield><subfield code="4">aut</subfield></datafield><datafield tag="700" ind1="1" ind2=" "><subfield code="a">Jiang, Shuihan</subfield><subfield code="4">aut</subfield></datafield><datafield tag="700" ind1="1" ind2=" "><subfield code="a">Qi, Junting</subfield><subfield code="4">aut</subfield></datafield><datafield tag="773" ind1="0" ind2="8"><subfield code="i">Enthalten in</subfield><subfield code="t">Multimedia tools and applications</subfield><subfield code="d">Springer US, 1995</subfield><subfield code="g">80(2020), 1 vom: 04. Sept., Seite 737-752</subfield><subfield code="w">(DE-627)189064145</subfield><subfield code="w">(DE-600)1287642-2</subfield><subfield code="w">(DE-576)052842126</subfield><subfield code="x">1380-7501</subfield><subfield code="7">nnns</subfield></datafield><datafield tag="773" ind1="1" ind2="8"><subfield code="g">volume:80</subfield><subfield code="g">year:2020</subfield><subfield code="g">number:1</subfield><subfield code="g">day:04</subfield><subfield code="g">month:09</subfield><subfield code="g">pages:737-752</subfield></datafield><datafield tag="856" ind1="4" ind2="1"><subfield code="u">https://doi.org/10.1007/s11042-020-09483-4</subfield><subfield code="z">lizenzpflichtig</subfield><subfield code="3">Volltext</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_USEFLAG_A</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">SYSFLAG_A</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_OLC</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">SSG-OLC-MAT</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">SSG-OLC-BUB</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">SSG-OLC-MKW</subfield></datafield><datafield tag="951" ind1=" " ind2=" "><subfield code="a">AR</subfield></datafield><datafield tag="952" ind1=" " ind2=" "><subfield code="d">80</subfield><subfield code="j">2020</subfield><subfield code="e">1</subfield><subfield code="b">04</subfield><subfield code="c">09</subfield><subfield code="h">737-752</subfield></datafield></record></collection>
|
author |
Zhang, Xiaojuan |
spellingShingle |
Zhang, Xiaojuan ddc 070 misc Object detection misc SSD misc Bounding box misc Center map misc Scale map Single shot object detection with refined feature |
authorStr |
Zhang, Xiaojuan |
ppnlink_with_tag_str_mv |
@@773@@(DE-627)189064145 |
format |
Article |
dewey-ones |
070 - News media, journalism & publishing 004 - Data processing & computer science |
delete_txt_mv |
keep |
author_role |
aut aut aut aut aut |
collection |
OLC |
remote_str |
false |
illustrated |
Not Illustrated |
issn |
1380-7501 |
topic_title |
070 004 VZ Single shot object detection with refined feature Object detection SSD Bounding box Center map Scale map |
topic |
ddc 070 misc Object detection misc SSD misc Bounding box misc Center map misc Scale map |
topic_unstemmed |
ddc 070 misc Object detection misc SSD misc Bounding box misc Center map misc Scale map |
topic_browse |
ddc 070 misc Object detection misc SSD misc Bounding box misc Center map misc Scale map |
format_facet |
Aufsätze Gedruckte Aufsätze |
format_main_str_mv |
Text Zeitschrift/Artikel |
carriertype_str_mv |
nc |
hierarchy_parent_title |
Multimedia tools and applications |
hierarchy_parent_id |
189064145 |
dewey-tens |
070 - News media, journalism & publishing 000 - Computer science, knowledge & systems |
hierarchy_top_title |
Multimedia tools and applications |
isfreeaccess_txt |
false |
familylinks_str_mv |
(DE-627)189064145 (DE-600)1287642-2 (DE-576)052842126 |
title |
Single shot object detection with refined feature |
ctrlnum |
(DE-627)OLC2122573473 (DE-He213)s11042-020-09483-4-p |
title_full |
Single shot object detection with refined feature |
author_sort |
Zhang, Xiaojuan |
journal |
Multimedia tools and applications |
journalStr |
Multimedia tools and applications |
lang_code |
eng |
isOA_bool |
false |
dewey-hundreds |
000 - Computer science, information & general works |
recordtype |
marc |
publishDateSort |
2020 |
contenttype_str_mv |
txt |
container_start_page |
737 |
author_browse |
Zhang, Xiaojuan Wang, Changying Cheng, Li Jiang, Shuihan Qi, Junting |
container_volume |
80 |
class |
070 004 VZ |
format_se |
Aufsätze |
author-letter |
Zhang, Xiaojuan |
doi_str_mv |
10.1007/s11042-020-09483-4 |
normlink |
(ORCID)0000-0002-2194-3257 |
normlink_prefix_str_mv |
(orcid)0000-0002-2194-3257 |
dewey-full |
070 004 |
title_sort |
single shot object detection with refined feature |
title_auth |
Single shot object detection with refined feature |
abstract |
Abstract Object classification and localization are two significant aspects of object detector based on the Single Shot MultiBox Detector (SSD). In general, the more feature maps there are, the better the object classification performance will be. However, when the information of excessive feature maps are sparse and unnecessary, the performance of object detection is slightly improved or maybe precisely opposite, which is instead harmful to the production of object localization. The performance of object detectors is not only related to the number of feature maps but also relies partly on the bounding box regression and Non-Maximum Suppression (NMS). In this paper, a detector is constructed based on SSD, called Detection with Refined Feature (DRF), involving center map and scale map, the detection loss is reshaped. Our motivation is to improve the accuracy of classification and localization by searching for central points and predicting the scales of the object points. Center map is used to predict the Intersection over Union (IoU) between the prediction box and ground truth box, while scale map considers the relationships among the different scales. Experimental results on both Pascal VOC and MS COCO 2014 instance datasets demonstrate the effectiveness of DRF. Using Darknet53, we achieve an 86.4% mean Average Precision (mAP) on Pascal VOC2007 and an 87.4% mAP on Pascal VOC2007 and VOC2012. On MS COCO, the DRF with ResNet50 still achieves moderate improvement. © Springer Science+Business Media, LLC, part of Springer Nature 2020 |
abstractGer |
Abstract Object classification and localization are two significant aspects of object detector based on the Single Shot MultiBox Detector (SSD). In general, the more feature maps there are, the better the object classification performance will be. However, when the information of excessive feature maps are sparse and unnecessary, the performance of object detection is slightly improved or maybe precisely opposite, which is instead harmful to the production of object localization. The performance of object detectors is not only related to the number of feature maps but also relies partly on the bounding box regression and Non-Maximum Suppression (NMS). In this paper, a detector is constructed based on SSD, called Detection with Refined Feature (DRF), involving center map and scale map, the detection loss is reshaped. Our motivation is to improve the accuracy of classification and localization by searching for central points and predicting the scales of the object points. Center map is used to predict the Intersection over Union (IoU) between the prediction box and ground truth box, while scale map considers the relationships among the different scales. Experimental results on both Pascal VOC and MS COCO 2014 instance datasets demonstrate the effectiveness of DRF. Using Darknet53, we achieve an 86.4% mean Average Precision (mAP) on Pascal VOC2007 and an 87.4% mAP on Pascal VOC2007 and VOC2012. On MS COCO, the DRF with ResNet50 still achieves moderate improvement. © Springer Science+Business Media, LLC, part of Springer Nature 2020 |
abstract_unstemmed |
Abstract Object classification and localization are two significant aspects of object detector based on the Single Shot MultiBox Detector (SSD). In general, the more feature maps there are, the better the object classification performance will be. However, when the information of excessive feature maps are sparse and unnecessary, the performance of object detection is slightly improved or maybe precisely opposite, which is instead harmful to the production of object localization. The performance of object detectors is not only related to the number of feature maps but also relies partly on the bounding box regression and Non-Maximum Suppression (NMS). In this paper, a detector is constructed based on SSD, called Detection with Refined Feature (DRF), involving center map and scale map, the detection loss is reshaped. Our motivation is to improve the accuracy of classification and localization by searching for central points and predicting the scales of the object points. Center map is used to predict the Intersection over Union (IoU) between the prediction box and ground truth box, while scale map considers the relationships among the different scales. Experimental results on both Pascal VOC and MS COCO 2014 instance datasets demonstrate the effectiveness of DRF. Using Darknet53, we achieve an 86.4% mean Average Precision (mAP) on Pascal VOC2007 and an 87.4% mAP on Pascal VOC2007 and VOC2012. On MS COCO, the DRF with ResNet50 still achieves moderate improvement. © Springer Science+Business Media, LLC, part of Springer Nature 2020 |
collection_details |
GBV_USEFLAG_A SYSFLAG_A GBV_OLC SSG-OLC-MAT SSG-OLC-BUB SSG-OLC-MKW |
container_issue |
1 |
title_short |
Single shot object detection with refined feature |
url |
https://doi.org/10.1007/s11042-020-09483-4 |
remote_bool |
false |
author2 |
Wang, Changying Cheng, Li Jiang, Shuihan Qi, Junting |
author2Str |
Wang, Changying Cheng, Li Jiang, Shuihan Qi, Junting |
ppnlink |
189064145 |
mediatype_str_mv |
n |
isOA_txt |
false |
hochschulschrift_bool |
false |
doi_str |
10.1007/s11042-020-09483-4 |
up_date |
2024-07-03T13:34:08.418Z |
_version_ |
1803565031403552768 |
fullrecord_marcxml |
<?xml version="1.0" encoding="UTF-8"?><collection xmlns="http://www.loc.gov/MARC21/slim"><record><leader>01000naa a22002652 4500</leader><controlfield tag="001">OLC2122573473</controlfield><controlfield tag="003">DE-627</controlfield><controlfield tag="005">20230505063909.0</controlfield><controlfield tag="007">tu</controlfield><controlfield tag="008">230505s2020 xx ||||| 00| ||eng c</controlfield><datafield tag="024" ind1="7" ind2=" "><subfield code="a">10.1007/s11042-020-09483-4</subfield><subfield code="2">doi</subfield></datafield><datafield tag="035" ind1=" " ind2=" "><subfield code="a">(DE-627)OLC2122573473</subfield></datafield><datafield tag="035" ind1=" " ind2=" "><subfield code="a">(DE-He213)s11042-020-09483-4-p</subfield></datafield><datafield tag="040" ind1=" " ind2=" "><subfield code="a">DE-627</subfield><subfield code="b">ger</subfield><subfield code="c">DE-627</subfield><subfield code="e">rakwb</subfield></datafield><datafield tag="041" ind1=" " ind2=" "><subfield code="a">eng</subfield></datafield><datafield tag="082" ind1="0" ind2="4"><subfield code="a">070</subfield><subfield code="a">004</subfield><subfield code="q">VZ</subfield></datafield><datafield tag="100" ind1="1" ind2=" "><subfield code="a">Zhang, Xiaojuan</subfield><subfield code="e">verfasserin</subfield><subfield code="0">(orcid)0000-0002-2194-3257</subfield><subfield code="4">aut</subfield></datafield><datafield tag="245" ind1="1" ind2="0"><subfield code="a">Single shot object detection with refined feature</subfield></datafield><datafield tag="264" ind1=" " ind2="1"><subfield code="c">2020</subfield></datafield><datafield tag="336" ind1=" " ind2=" "><subfield code="a">Text</subfield><subfield code="b">txt</subfield><subfield code="2">rdacontent</subfield></datafield><datafield tag="337" ind1=" " ind2=" "><subfield code="a">ohne Hilfsmittel zu benutzen</subfield><subfield code="b">n</subfield><subfield code="2">rdamedia</subfield></datafield><datafield tag="338" ind1=" " ind2=" "><subfield code="a">Band</subfield><subfield code="b">nc</subfield><subfield code="2">rdacarrier</subfield></datafield><datafield tag="500" ind1=" " ind2=" "><subfield code="a">© Springer Science+Business Media, LLC, part of Springer Nature 2020</subfield></datafield><datafield tag="520" ind1=" " ind2=" "><subfield code="a">Abstract Object classification and localization are two significant aspects of object detector based on the Single Shot MultiBox Detector (SSD). In general, the more feature maps there are, the better the object classification performance will be. However, when the information of excessive feature maps are sparse and unnecessary, the performance of object detection is slightly improved or maybe precisely opposite, which is instead harmful to the production of object localization. The performance of object detectors is not only related to the number of feature maps but also relies partly on the bounding box regression and Non-Maximum Suppression (NMS). In this paper, a detector is constructed based on SSD, called Detection with Refined Feature (DRF), involving center map and scale map, the detection loss is reshaped. Our motivation is to improve the accuracy of classification and localization by searching for central points and predicting the scales of the object points. Center map is used to predict the Intersection over Union (IoU) between the prediction box and ground truth box, while scale map considers the relationships among the different scales. Experimental results on both Pascal VOC and MS COCO 2014 instance datasets demonstrate the effectiveness of DRF. Using Darknet53, we achieve an 86.4% mean Average Precision (mAP) on Pascal VOC2007 and an 87.4% mAP on Pascal VOC2007 and VOC2012. On MS COCO, the DRF with ResNet50 still achieves moderate improvement.</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Object detection</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">SSD</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Bounding box</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Center map</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Scale map</subfield></datafield><datafield tag="700" ind1="1" ind2=" "><subfield code="a">Wang, Changying</subfield><subfield code="4">aut</subfield></datafield><datafield tag="700" ind1="1" ind2=" "><subfield code="a">Cheng, Li</subfield><subfield code="4">aut</subfield></datafield><datafield tag="700" ind1="1" ind2=" "><subfield code="a">Jiang, Shuihan</subfield><subfield code="4">aut</subfield></datafield><datafield tag="700" ind1="1" ind2=" "><subfield code="a">Qi, Junting</subfield><subfield code="4">aut</subfield></datafield><datafield tag="773" ind1="0" ind2="8"><subfield code="i">Enthalten in</subfield><subfield code="t">Multimedia tools and applications</subfield><subfield code="d">Springer US, 1995</subfield><subfield code="g">80(2020), 1 vom: 04. Sept., Seite 737-752</subfield><subfield code="w">(DE-627)189064145</subfield><subfield code="w">(DE-600)1287642-2</subfield><subfield code="w">(DE-576)052842126</subfield><subfield code="x">1380-7501</subfield><subfield code="7">nnns</subfield></datafield><datafield tag="773" ind1="1" ind2="8"><subfield code="g">volume:80</subfield><subfield code="g">year:2020</subfield><subfield code="g">number:1</subfield><subfield code="g">day:04</subfield><subfield code="g">month:09</subfield><subfield code="g">pages:737-752</subfield></datafield><datafield tag="856" ind1="4" ind2="1"><subfield code="u">https://doi.org/10.1007/s11042-020-09483-4</subfield><subfield code="z">lizenzpflichtig</subfield><subfield code="3">Volltext</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_USEFLAG_A</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">SYSFLAG_A</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_OLC</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">SSG-OLC-MAT</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">SSG-OLC-BUB</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">SSG-OLC-MKW</subfield></datafield><datafield tag="951" ind1=" " ind2=" "><subfield code="a">AR</subfield></datafield><datafield tag="952" ind1=" " ind2=" "><subfield code="d">80</subfield><subfield code="j">2020</subfield><subfield code="e">1</subfield><subfield code="b">04</subfield><subfield code="c">09</subfield><subfield code="h">737-752</subfield></datafield></record></collection>
|
score |
7.4007874 |