Useful Links related to GRDDC’2020
GitHub Website for latest news and updates
ResearchGate Project for the latest publication updates
GRDDC Summary Paper – IEEE Conference Proceedings
GRDDC Summary Paper – Preprint – Open Access
GRDDC Workshop – Challenge Overview Video
GRDDC Workshop – Image Gallery()
Data Article for RDD2020 (Open Access)
RDD2020 Dataset on Mendeley
Other Related Articles( Article 1: RDD2020 , Article 2: RDD2019 , Article 3: RDD2018 )
GRDDC’2020: Concluding Remarks
IEEE Big Data Cup – GRDDC’2020 culminated successfully in December 2020. The concluding remarks and the details of participants and winners are provided below.
121 teams from different nations registered for the challenge. 13 teams were shortlisted based on the average F1-score achieved by the proposed solutions and the submissions of source-code. Out of these, 12 teams made it to the final round requiring a detailed report of their proposed solution. The details of the teams and the solutions proposed are provided below.
The top 12 teams were given an opportunity to present their solutions in the Big Data Cup workshop conducted in association with the IEEE International Conference on Big Data 2020, GA, USA , on December 10, 2020. The proposed solutions were published as research articles in the conference proceedings.
he top three teams were awarded Cash-Prize ($1500, $1000, and $500) along with free registration to the conference.
Table 1 lists the details of ranking and contributors of the top 12 teams along with their scores corresponding to the two leader-boards. A brief overview of the solutions proposed by these teams is presented in the next table. The further details about the solutions and the source codes can be accessed in the GRDDC summary paper, here.
|
GRDDC – Rank |
Name of the Team |
Contributors |
Test1- Score |
Test2- Score |
|
1 |
IMSC |
Vinuta Hegde, Dweep Trivedi, Abdullah Alfarrarjeh, Aditi Deepak, Seon Ho Kim, Cyrus Shahabi |
0.6748 |
0.6662 |
|
2 |
SIS Lab |
Keval Doshi, Yasin Yilmaz |
0.6275 |
0.6358 |
|
3 |
DD- VISION |
Zixiang Pei, Xiubao Zhang, Rongheng Lin, Haifeng Shen, Jian Tang, and Yi Yang |
0.629 |
0.6219 |
|
4 |
titan_mu |
Vishal Mandal, Yaw Adu-Gyamfi, Abdul Rashid Mussah |
0.5814 |
0.5751 |
|
5 |
Dongjuns |
Dongjun Jeong |
0.5683 |
0.5710 |
|
6 |
SUTPC |
Yuming Liu, Xiaoyong Zhang, Bingzhen Zhang, and Zhenwu Chen |
0.5636 |
0.5707 |
|
7 |
RICS |
Sadra Naddaf-Sh, M-Mahdi Naddaf-Sh, Amir R. Kashani, Hassan Zargarzadeh |
0.565 |
0.547 |
|
8 |
AIRS- CSR |
Xiaoguang Zhang, Xuan Xia, Nan Li, Ma Lin, Junlin Song, Ning Ding |
0.554 |
0.541 |
|
9 |
CS17 |
Tristan Hascoet, Yihao Zhang, Andreas Persch, Ryoichi Takashima, Tetsuya Takiguchi, and Yasuo Ariki |
0.5413 |
0.5430 |
|
10 |
BDASL |
Rahul Vishwakarma, Ravigopal Vennelakanti |
0.5368 |
0.5426 |
|
11 |
IDVL |
Vung Pham, Chau Pham, Tommy Dang |
0.51 |
0.514 |
|
12 |
E-LAB |
Felix Kortmann, Kevin Talits, Pascal Fassmeyer, Alexander Warnecke, Nicolas Meier, Jens Heger, Paul Drews, Burkhardt Funk |
0.4720 |
0.4656 |
Team Name |
Proposed Solution |
Ensemble Learning? |
Data Augmentation |
IMSC |
Ensemble Learning with u-YOLO and Test Time Augmentation |
Yes |
Yes |
SIS Lab |
Ensemble model with YOLO-v4 as base model. |
Yes |
Yes |
DD- VISION |
A Consistency Filtering Mechanism and model ensemble with cascade R-CNN as the base model |
Yes |
Yes |
titan_mu |
YOLO model trained on CSPDarknet53 backbone |
No |
Not used |
Dongjuns |
YOLOv5x |
No |
Yes |
SUTPC |
Ensemble(YOLO-v4 and Faster-RCNN) |
Yes |
Yes |
RICS |
EfficientDet |
No |
Explored but not used |
AIRS- CSR |
YOLOv4 |
No |
Yes |
CS17 |
Resnet-101 backbone based Faster-RCNN two-stage detection architecture |
No |
Explored but not used |
BDASL |
Multi-stage Faster R-CNN with Resnet-50 and Resnet-101 backbones |
No |
Yes |
IDVL |
Road Damage Detector using Detectron2 and Faster R-CNN |
No |
Yes |
E-LAB |
FR-CNN; Classifying the region and using regional experts for the detection |
No |
Yes |
We are delighted to have the privilege of interacting and working with the tech enthusiasts of top notch organisations and institutes, and applaud the hard work and progress of all the participating teams and remind everyone that it’s just the beginning and we will emerge better, stronger and more successful in the upcoming years.
After the challenge submissions were closed and the leader-boards were frozen, we received several requests to re-open the leader-boards. To address these requests new leader-boards have been created on the challenge website.
The researchers may sign-up for free , perform experiments, submit their solutions - predicted annotations for test1 and test2 images (for details – please refer the guidelines on the website, and the GRDDC Summary Paper ) and evaluate using the new leader-boards.
If you use the resources (data, content, leader-boards etc.), please cite the following:
Arya, D., Maeda, H., Ghosh, S. K., Toshniwal, D., & Sekimoto, Y. (2021). RDD2020: An annotated image dataset for Automatic Road Damage Detection using Deep Learning. Data in Brief, 107133.
[dataset] Arya, D., Maeda, H., Ghosh, S. K., Toshniwal, D., Omata, H., Kashiyama, T., Seto, T., Mraz, A., & Sekimoto, Y. (2021), “RDD2020: An Image Dataset for Smartphone-based Road Damage Detection and Classification”, Mendeley Data, V1, doi: 10.17632/5ty2wb6gvg.1
Arya, D., Maeda, H., Ghosh, S. K., Toshniwal, D., Omata, H., Kashiyama, T., & Sekimoto,
Y. (2020). Global Road Damage Detection: State-of-the-art Solutions. IEEE International Conference on Big Data (Big Data), Atlanta, GA, USA, 2020, pp. 5533-5539, doi: 10.1109/BigData50022.2020.9377790.
Arya, D., Maeda, H., Ghosh, S. K., Toshniwal, D., Mraz, A., Kashiyama, T., & Sekimoto, Y. (2020). Transfer Learning-based Road Damage Detection for Multiple Countries. arXiv preprint arXiv:2008.13101.
Maeda, H., Kashiyama, T., Sekimoto, Y., Seto, T., & Omata, H. (2020) Generative adversarial network for road damage detection. Computer‐Aided Civil and Infrastructure Engineering.