Iteration Plan - Oct 2020

[dreamerlin]

MMAction2 v0.7 is released. Now it's time to plan for the next iteration, i.e. v0.8.
This iteration will focus on supporting Spatio-Temporal Action Detction, multi-modality training and enriching algorithms/models for action recognition.

Algorithms:

• X3D (Add X3d models #288)
• C3D ([Feature] Support C3D #226)

Datasets:

• HVU (HVU baselines #287)
• AVA ([Feature] Add AVA preparation #238, [Feature] Add AVA dataset and transformation #266)

Tools:

• Omni-source pipeline ([Feature] Support Omnisource Dataset #294, [Feature] Add Omnisource training configs #296)
• Audio dataset pipeline ([Feature] MultiModality: Audio #205)
• Text-embedding dataset pipeline
• Multi-modality fusion results for Kinetics-400 (Update the audio fusion result with a better tsn-f3 result. #281)

Misc:

---

Deferred

• SlowFast on AVA
• MOC