Recognize more people and Understand more questions - Gin-TrungSon/EllieHumanoid GitHub Wiki

Face Dataset

First of all, you need to add faces to face dataset. Here ist the folder of dataset. You just need to create a separate sub directory that contains pictures of a person. It's enough if you have around 4-5 pictures for each person.

- data
  - person1
    - img01.jpg
    - img02.jpg
    - img03.jpg
  - person2
    - img01.jpg
    - img02.jpg
    - img03.jpg

Voice assistant intents

If you want to add more questions and answers to help Ellie smarter, then put your intents at this file. For example :

{"intents": [
    {"tag": "greeting", # The identity of the intent
     "patterns": ["Grüß Gott!", "Guten Tag"], # The patterns for recognizing this intent
     "responses": ["Grüß Gott!"," was kann ich für Sie tun"], # The answers for this intent
    },
    {"tag": "goodbye",
     "patterns": ["Schönen Abend noch", "auf Wiederschauen" ]
    },
]
}

After that just run this python file to train a new model. Training with a power computer is recommended and then we can replace files "intents/intents.json" ,"weights/ellie.dts" and "data/data.txt" with new trained files.