How to use

Do it your self

If you want to use your own videos from locally or online sources follow the following steps

In case you want to insert Local videos, go to this folder and insert your videos there.
In case you want to insert Online videos, go to this file and insert the youtube link video there.

For these modules it's just needed to run all the cells in the jupyter notebooks located in each folder.

At this module you can choose which video you want to use as the seed. If you want to insert others seeds, is only necessary to insert than into this folder.
With the videos you want as seeds in the folder, open the following notebooks and run all the code cells in it
After running the notebooks, open this last notebook, go to the seventh code cell and change the VIDEO_ID variable for the video ID you want to create the seed. (To check the VIDEO_ID of your video just look at the table created by the previous code cell which has all the reference videos with it's id's names and other information, here an example of this table).
Go to this file, at the Select the starting and ending frame to mark cell (tenth cell) and change the init frame and end frame to fit your video specifications. After go to the thirteenth cell and select the class, and the measures you want to mark the video as. After just run the left cells.

For this module you need to go to this notebook, and run all the cells in it if you want to use all your seed videos to mark all your dataset videos. Make sure to select the frame_distance_threshold variable and the euclidean_distance_threshold variable as your preference (or you can just keep the values that are already in the notebook). The first one is responsible for the delay in which a series can be considered, x axis, decreasing this variable can impact in a lower number of matching founds. The second one is the distance of the euclidean matching algorithm that will be considered, decreasing this variable can impact in a bigger number of matching founds.
After that, will be created a VD_LABELED_L0.CSV file in every video folder with the occurrences of the seeds you searched for.
To check if the seed is present in a video just go to VD_LABELED_L0.CSV file of your video, at the last column on the right and scroll down, if the seed was marked, it will appear like in the Figure 1 at the frames it was found.

Figure 1 - Example of labeled frame.

For the neural network process, it's necessary to first prepare the data, for this step is only necessary to run all code cells in prepare data notebook

For training your model go to the training notebook and just run all the cells in it. The result is going to be a model saved at the folder of the notebook. You can also find the training and evaluation process in the same notebook.
If you want to understand or check the model developed in this research as well the results we achieved using this model, check at this section of the wiki.