Papers of the day   All papers

EfficientDet: Scalable and Efficient Object Detection

Comments

Quoc Le: EfficientDet: a new family of efficient object detectors. It is based on EfficientNet, and many times more efficient than state of art models. Link: https://arxiv.org/abs/1911.09070 Code: coming soon https://t.co/2KYabAnpLL

9 replies, 822 likes


Mingxing Tan: Happy to share EfficientDet-D7x: new SOTA 55.1 COCO AP at 153ms latency (10x faster than other multi-scale models). TF2+keras supported! Arxiv: https://arxiv.org/abs/1911.09070 github: https://bit.ly/307N6GG paperwithcode: https://bit.ly/2Esz89U https://t.co/swvnORMTRK

7 replies, 470 likes


Tomasz Malisiewicz: EfficientDet: Scalable and Efficient Object Detection. A new family of object detectors that achieves an order-of-magnitude better efficiency than prior art. +Large scale experiments from Google Brain https://arxiv.org/abs/1911.09070 #computervision #Robotics https://t.co/9mflLs3q0b

5 replies, 446 likes


Mingxing Tan: Excited to announce the open source of EfficientDet: better accuracy & efficiency on COCO detection. Bonus: it also works pretty well for semantic segmentation (Table 3)! Paper: https://arxiv.org/abs/1911.09070 Code: https://github.com/google/automl/tree/master/efficientdet https://t.co/pV2EfnQIVl

6 replies, 440 likes


Mingxing Tan: Excited to share our work on efficient neural architectures for object detection! New state-of-the-art accuracy (51 mAP on COCO for single-model single-scale), with an order-of-magnitude better efficiency! Collaborated with @quocleix and @ruomingpang.

2 replies, 130 likes


tdual🍮MatrixFlow: 2018 11/12 M2Det https://arxiv.org/abs/1811.04533 2019 7/11 A Survey of Deep Learning-based Object Detection https://arxiv.org/abs/1907.09408 2019 11/20 EfficientDet https://arxiv.org/abs/1911.09070 2020 4/23 YOLOv4 https://arxiv.org/abs/2004.10934

2 replies, 65 likes


Aakash Kumar Nain: I will be writing a blog post explaining this amazing paper but only after next week. I am on a break right now.

1 replies, 55 likes


hiroto: EfficientDet has two new features: EfficientNet backbone and BiFPN. It's quite helpful the paper shows the separate contribution of them. Res50 + FPN (standard RetinaNet) : 37.0 mAP EfficientNet B3 + FPN : 40.3 mAP from: https://arxiv.org/abs/1911.09070

1 replies, 42 likes


phalanx: EfficientDet: Scalable and Efficient Object Detection https://arxiv.org/abs/1911.09070 https://t.co/WHUT6E6VXl

0 replies, 28 likes


Mingxing Tan: Excited to see self-training obtains SoTA accuracy on COCO detection and Pascal segmentation. What if you also need efficiency? Try out our updated EfficientDet (53.7AP, with 55M params and 122ms latency): https://arxiv.org/abs/1911.09070. Enjoy :)

0 replies, 24 likes


Brandon Rohrer: I love the emphasis on efficiency, simplification, and multiscale features in this approach. Smaller is beautifuller, and state of the art performance is gravy.

0 replies, 20 likes


Karol Majek, PhD: EfficientDet (red one) is much faster and better than other SoA object detection nets! https://arxiv.org/abs/1911.09070 https://t.co/B9imj71bE4

0 replies, 15 likes


Carlo Lepelaars: Reading the Noisy Student and EfficientDet papers. @quocleix and the other researchers made a big breakthrough with EfficientNet and now we are reaping the benefits of these more efficient models. 😎 Noisy Student: https://arxiv.org/pdf/1911.04252.pdf EfficientDet: https://arxiv.org/pdf/1911.09070.pdf

1 replies, 12 likes


Bojan Tunguz: EfficientDet: Scalable and Efficient Object Detection https://arxiv.org/pdf/1911.09070.pdf

0 replies, 11 likes


Andrew Davison: Looks worth checking out... More efficient object detection.

0 replies, 11 likes


akira: https://arxiv.org/abs/1911.09070 High-speed and high-accuracy object detection network, EfficientDet ,with search space like EfficientNet and BiFPN, kind of Feature Pyramid Network that flows information from both high and low resolutions. It achieve SOTA but about 10 times faster. https://t.co/LqpISudTm7

0 replies, 6 likes


arXiv CS-CV: EfficientDet: Scalable and Efficient Object Detection http://arxiv.org/abs/1911.09070

0 replies, 4 likes


Ajay Tanwani: Using bidirectional feature pyramid network for object detection with surprisingly faster convergence

0 replies, 4 likes


arXiv CS-CV: EfficientDet: Scalable and Efficient Object Detection http://arxiv.org/abs/1911.09070

0 replies, 3 likes


arXiv CS-CV: EfficientDet: Scalable and Efficient Object Detection http://arxiv.org/abs/1911.09070

0 replies, 3 likes


Daisuke Okanohara: EfficientDet achieves efficient and accurate object detection by 1) using EfficientNet as a backbone 2) BiFPN based on PANet 3) scaling depth, channel, parameters jointly (as EfficientNet). Current networks seem still far from the optimum yet. https://arxiv.org/abs/1911.09070

0 replies, 3 likes


Brandon Rohrer: Exhibit A is Figure 3 from "EfficientDet: Scalable and Efficient Object Detection" by @tanmingxing @ruomingpang and @quocleix https://arxiv.org/pdf/1911.09070.pdf

1 replies, 3 likes


Rosinality: https://arxiv.org/abs/1911.09070 https://t.co/F9eLwILgLc

1 replies, 2 likes


Popular ML resources: The most popular ArXiv tweet in the last 24h: https://twitter.com/tanmingxing/status/1239959559969488897

0 replies, 1 likes


Umberto Michelucci: Have to read the paper...

0 replies, 1 likes


Brundage Bot: EfficientDet: Scalable and Efficient Object Detection. Mingxing Tan, Ruoming Pang, and Quoc V. Le http://arxiv.org/abs/1911.09070

1 replies, 0 likes


Content

Found on Nov 22 2019 at https://arxiv.org/pdf/1911.09070.pdf

PDF content of a computer science paper: EfficientDet: Scalable and Efficient Object Detection