# Elaine's Feedback

**Yick you can continue by writing is that ok? sorry**

* Justify why we use OpenPose
* Justify why we use LSTMs
* Papers actually run on different models and then choose the best one
* We don't have time - cite performances in Related Work/Lit Review
* Justify results of our model evaluation
*Given set of hardware configurations*
* Accuracy Metrics
* Latency Metrics
* Sensitivity testing

### Discussing effects of changes
* Given same model configuration
* Increasing number of signs
* Set a specific confidence values:
* Tracking number of datasets needed
* Number of epochs to reach
* Alternatively, increase number