In this task, participants are supposed to generate co-speech gesture videos for given audio.
To evaluate the quality of the generated  co-speech gesture videos, we use both objective metrics and subjective metrics.

###Objective metrics
- **Fréchet Video Distance** (**FVD**) [3] to measure the similarity between two video sequences. It takes into account both spatial and temporal relationships between the videos, allowing for a more accurate assessment of their similarity.
- **Fréchet Gesture Distance** (**FGD**) [4] to evaluate the distance between the real and synthetic gesture distribution. We train an auto-encoder on poses extracted from both the competition dataset and PATS dataset to get motion features. Fréchet distance between the real and synthetic gesture is calculated in motion feature space.
- **Beat Consistency Score** (**BC**) [5] to measure the speech-motion alignment. 
- **Diversity** (**Div.**) [6] to measure the diversity among generated gestures.

**Note 1**:  **FGD**, **BC** and **Div.**  *evaluate generated videos from a human motion perspective. We first use [MMPose](https://github.com/open-mmlab/mmpose/tree/main) to estimate poses of videos.*

###Subjective metrics
- **Realness** to evaluate the generation quality on the video level, whether the video is generated or captured. 
- **Human-likeness** to evaluate whether the generated gestures look natural and are difficult to distinguish from captured human motions.
- **Gesture-speech appropriateness** to evaluate whether the generated gestures are consistent with the input audio, in terms of both rhythm and semantics.
- **Diversity (Div.)** to evaluate whether the generated gestures are diverse enough.

**Note 2**: *The rating scale for the subjective metrics is 1 to 5, with 1 being the poorest and 5 being the best.*
**Note 3**: *The Leaderboard only shows the ranking of **FVD**.  Other metrics will only be evaluated on the final submission.*

The final ranking would be determined by the number of "first place" across all metrics. If multiple teams win the same number of "first place" metrics, the final ranking would be determined by the number of "second place" across all metrics, and so on. 

## Submission File
Your team ultimately needs to submit a compressed zip file(*.zip), for instance, "**submission.zip**".
The compressed zip file contains two parts: submission.csv and your generated videos.
```
|--submission.zip
|----submission.csv
|----PersonA_test.mp4
|----PersonB_test.mp4
```
For submission CSV files, you should use the "**extraction_FVD_to_submission.py**" (submission_form.zip from Dataset) tool to process your generated videos. 
**!!!!! Tips: Please ensure that the file name of the video you generate matches our requirements: PersonA_test.mp4, PersonB_test.mp4**
For each ID in the test set, you must generate a video for a given audio and the video should be named according to the following format "ID_test.mp4", for example, "PersonA_test.mp4".

##Reference
[1] Audio-Driven Co-Speech Gesture Video Generation, NeurIPS 2022: ANGIE https://github.com/alvinliu0/ANGIE.
[2] Audio-Driven Neural Gesture Reenactment with Video Motion Graphs, CVPR 2022:  vid-reenact https://github.com/yzhou359/vid-reenact.
[3] Towards accurate generative models of video: A new metric & challenges. arXiv:1812.01717, 2018.
[4] Learn to Dance with AIST++: Music Conditioned 3D Dance Generation, ICCV2021: https://github.com/google-research/mint
[5] Speech Gesture Generation from the Trimodal Context of Text, Audio, and Speaker Identity, TOG2020: https://github.com/ai4r/Gesture-Generation-from-Trimodal-Context#gesture-generation-from-trimodal-context
[6] Learning Hierarchical Cross-Modal Association for Co-Speech Gesture Generation, CVPR2022: HA2G https://github.com/alvinliu0/HA2G