The fourth phase has begun. You can now submit your Deepfake detection codes to the competition website. Please note that the valid submission slot of the fourth phase is from now to Apr. 3, 24:00. When you click the “Submit” button, you may need to wait for seconds or minutes before the submission is fully uploaded, depending on your network speed and submission size. After submission, the status “submitted” indicate that your submission is completed and waiting for evaluation, and the status “running” indicates that your submission is being evaluated on our server. If everything goes fine, your best result will be shown on the Leaderboard (if not, please manually submit your best score to LB). Also note that the “detailed results” in the LB is not working for now due to some bug of Codalab platform. However, you can still view your own detailed results by “Download output from scoring step” in your submission transaction and see the “scores.txt”. Also, please team up if you work as a team.
Before submission, it is recommended that you first test your codes using our starter-kit and using the specified docker environment.
Your detection submissions in this phase will be evaluated on the 10 datasets in the last creation phase, and AUC is used as the detection metric. Our baseline detection method (a Xception model) has the following detailed results:
Score:0.606769
ExecutionTime:86
Det_score_joshhu:0.014418
Det_score_seanseattle:0.493267
Det_score_zhaobh:0.495223
Det_score_jerryHUST:0.558894
Det_score_combaz:0.777828
Det_score_DFGC_SYSU:0.351905
Det_score_nodo:0.574626
Det_score_yZzzzzz:0.812093
Det_score_organizer_baseline:0.995284
Det_score_hlb01:0.994151
Note please fix any randomness in your detection codes. Possible randomness can come from using albumentation for image preprocessing, etc.. If we find any randomness in your detection codes, your submission may not be used in the following evaluation phase and may not be eligible for awards.
PS: Do NOT use CelebDF-v2 test set in your training or development, and do NOT use extra data other than CelebDF-v2 train set during training of detection phases. Do NOT use filenames, metadata, or facial ID information for detection. Follow the rules listed in “Terms and Conditions”.
Have fun!
DFGC Organizers
Dear friends, how should I solve this problem：
Execution time limit exceeded!
We have a runtime limitation of 5 mins. Please try to optimize your model and reduce execution time.
Posted by: bob_peng @ April 1, 2021, 1:14 p.m.