ForgeryNet: Face Forgery Analysis Challenge 2021

Organized by heyinan - Current server time: Oct. 20, 2021, 1:59 p.m. UTC

First phase

Track1: Forgery Image Analysis
July 12, 2021, midnight UTC

End

Competition Ends
Sept. 19, 2021, 11:59 p.m. UTC

Track1: Forgery Image Analysis

Start: July 12, 2021, midnight

Description: The online evaluation results must be submitted through this CodaLab competition site of the ForgeryNet Challenge. The online evaluation chances will be updated on Friday, 11:59 p.m (UTC) every week. The organizer will offer some hints/codes to ensure a successful submission. Note that challenge participants cannot use other external data, and the validation set cannot be used as training data.

Track2: Forgery Video Analysis

Start: July 12, 2021, midnight

Description: The online evaluation results must be submitted through this CodaLab competition site of the ForgeryNet Challenge. The online evaluation chances will be updated on Friday, 11:59 p.m (UTC) every week. The organizer will offer some hints/codes to ensure a successful submission. Note that challenge participants cannot use other external data, and the validation set cannot be used as training data.

Track3: Forgery Video Temporal Localization

Start: July 12, 2021, midnight

Description: The online evaluation results must be submitted through this CodaLab competition site of the ForgeryNet Challenge. The online evaluation chances will be updated on Friday, 11:59 p.m (UTC) every week. The organizer will offer some hints/codes to ensure a successful submission. Note that challenge participants cannot use other external data, and the validation set cannot be used as training data.

Private Test

Start: Sept. 12, 2021, midnight

Description: 1. Every submission should be contained in a docker image. [You should submit a public docker image path] 2. No internet access enabled. 2. docker <= 45 minutes run-time on 24 VCPU and Single Tesla V100-SXM2 32G. We recommend that you do not exceed the time limit. 3. External data must be freely & publicly available, including pre-trained models. Notice that any temp file when runing this container should be writen at `/forgerynet_output/$your_team_name`. You can create this folder in your `run.sh`. And DO NOT create `/forgerynet_output/` in your images. 4. The dataset is mounted in the `/forgerynet_data/` directory in the container in a read-only manner. 5. Output format should be same as public test. 6. You can pull `forgerynet/forgerynet-submit` in docker hub to get an example. 7. Send your docker image path to forgerynet@gmail.com, we will response you as soon as possible.

Competition Ends

Sept. 19, 2021, 11:59 p.m.

You must be logged in to participate in competitions.

Sign In