CornerNet-Lite: Efficient Keypoint Based Object Detection
H. Law, Y. Teng, O. Russakovsky, and J. Deng. (2019)cite arxiv:1904.08900Comment: Code available at https://github.com/princeton-vl/CornerNet-Lite.
Abstract
Keypoint-based methods are a relatively new paradigm in object detection,
eliminating the need for anchor boxes and offering a simplified detection
framework. Keypoint-based CornerNet achieves state of the art accuracy among
single-stage detectors. However, this accuracy comes at high processing cost.
In this work, we tackle the problem of efficient keypoint-based object
detection and introduce CornerNet-Lite. CornerNet-Lite is a combination of two
efficient variants of CornerNet: CornerNet-Saccade, which uses an attention
mechanism to eliminate the need for exhaustively processing all pixels of the
image, and CornerNet-Squeeze, which introduces a new compact backbone
architecture. Together these two variants address the two critical use cases in
efficient object detection: improving efficiency without sacrificing accuracy,
and improving accuracy at real-time efficiency. CornerNet-Saccade is suitable
for offline processing, improving the efficiency of CornerNet by 6.0x and the
AP by 1.0% on COCO. CornerNet-Squeeze is suitable for real-time detection,
improving both the efficiency and accuracy of the popular real-time detector
YOLOv3 (34.4% AP at 34ms for CornerNet-Squeeze compared to 33.0% AP at 39ms for
YOLOv3 on COCO). Together these contributions for the first time reveal the
potential of keypoint-based detection to be useful for applications requiring
processing efficiency.
Description
CornerNet-Lite: Efficient Keypoint Based Object Detection
%0 Generic
%1 law2019cornernetlite
%A Law, Hei
%A Teng, Yun
%A Russakovsky, Olga
%A Deng, Jia
%D 2019
%K backbone code cornernet detection head
%T CornerNet-Lite: Efficient Keypoint Based Object Detection
%U http://arxiv.org/abs/1904.08900
%X Keypoint-based methods are a relatively new paradigm in object detection,
eliminating the need for anchor boxes and offering a simplified detection
framework. Keypoint-based CornerNet achieves state of the art accuracy among
single-stage detectors. However, this accuracy comes at high processing cost.
In this work, we tackle the problem of efficient keypoint-based object
detection and introduce CornerNet-Lite. CornerNet-Lite is a combination of two
efficient variants of CornerNet: CornerNet-Saccade, which uses an attention
mechanism to eliminate the need for exhaustively processing all pixels of the
image, and CornerNet-Squeeze, which introduces a new compact backbone
architecture. Together these two variants address the two critical use cases in
efficient object detection: improving efficiency without sacrificing accuracy,
and improving accuracy at real-time efficiency. CornerNet-Saccade is suitable
for offline processing, improving the efficiency of CornerNet by 6.0x and the
AP by 1.0% on COCO. CornerNet-Squeeze is suitable for real-time detection,
improving both the efficiency and accuracy of the popular real-time detector
YOLOv3 (34.4% AP at 34ms for CornerNet-Squeeze compared to 33.0% AP at 39ms for
YOLOv3 on COCO). Together these contributions for the first time reveal the
potential of keypoint-based detection to be useful for applications requiring
processing efficiency.
@misc{law2019cornernetlite,
abstract = {Keypoint-based methods are a relatively new paradigm in object detection,
eliminating the need for anchor boxes and offering a simplified detection
framework. Keypoint-based CornerNet achieves state of the art accuracy among
single-stage detectors. However, this accuracy comes at high processing cost.
In this work, we tackle the problem of efficient keypoint-based object
detection and introduce CornerNet-Lite. CornerNet-Lite is a combination of two
efficient variants of CornerNet: CornerNet-Saccade, which uses an attention
mechanism to eliminate the need for exhaustively processing all pixels of the
image, and CornerNet-Squeeze, which introduces a new compact backbone
architecture. Together these two variants address the two critical use cases in
efficient object detection: improving efficiency without sacrificing accuracy,
and improving accuracy at real-time efficiency. CornerNet-Saccade is suitable
for offline processing, improving the efficiency of CornerNet by 6.0x and the
AP by 1.0% on COCO. CornerNet-Squeeze is suitable for real-time detection,
improving both the efficiency and accuracy of the popular real-time detector
YOLOv3 (34.4% AP at 34ms for CornerNet-Squeeze compared to 33.0% AP at 39ms for
YOLOv3 on COCO). Together these contributions for the first time reveal the
potential of keypoint-based detection to be useful for applications requiring
processing efficiency.},
added-at = {2019-05-12T22:33:08.000+0200},
author = {Law, Hei and Teng, Yun and Russakovsky, Olga and Deng, Jia},
biburl = {https://www.bibsonomy.org/bibtex/2c6d99d46c5260c21ad74d2a2cd3379ef/nmatsuk},
description = {CornerNet-Lite: Efficient Keypoint Based Object Detection},
interhash = {37cabe83eb3f9677cdf2b97d4285f304},
intrahash = {c6d99d46c5260c21ad74d2a2cd3379ef},
keywords = {backbone code cornernet detection head},
note = {cite arxiv:1904.08900Comment: Code available at https://github.com/princeton-vl/CornerNet-Lite},
timestamp = {2019-05-12T22:33:08.000+0200},
title = {CornerNet-Lite: Efficient Keypoint Based Object Detection},
url = {http://arxiv.org/abs/1904.08900},
year = 2019
}