Week 1. Record videos to use prediction models

Recording the videos

Now that everything is clear about how to generate the correct dataset, which are the models that work better for this type of predictions and how to get a complete Neural Network model is time to start working with images recorded by me and see how the models fit with them.

Then first I need some home made videos with a static background and a ball going from the beggining to the end. (From left to right) The idea is to have a similarity between the datasets generated and the videos I recorded, because it is the only way we can use this training models and approach a result in real life.

So this week my task is to record around 150 videos for use them in next steps to test our Neural Networks. The better option is to have so many videos recorded and use them for training the model, not just to test, but it is a very hard task to get 20000 samples, we can modify the aspect in our dataset to match the caracteristics in the videos recorded.

This is and example of the videos I recorded: