This was very informative and enlightening. It hits on different level when you see something working like this, just got so many implementation ideas soo early in my journey of learning, just by this video. High on curiosity. 🧨

Just a glimpse of Instructor being used with LLaMa https://python.useinstructor.com/hub/llama-cpp-python/#llama-cpp-python This proves how powerful Instructor is.

@Dai Nguyen According to me 1. You can search for all the models available and compare them by going through their documentation. 2. Select the best model catering to your need 3. Fine tune the model according to your requirements, beware of overfitting. 4. Test if it is working as expected. For MediaPipe you can follow the below steps: 1. You can start by forking this repo https://github.com/google/mediapipe 2. Go through all the readme’s 3. Go through the release notes 4. Clone it in your system 5. Try running it without training first 6. If the results are satisfying voila, otherwise fine tune it according to your need. Beware of overfitting. PS: Here is the link which might help you to pick a best fit model for you need https://pallawi-ds.medium.com/which-human-pose-estimation-model-should-you-pick-to-realise-your-ideas-for-a-video-analytics-6ca754cc1f4e Happy Learning 🧨