Computer Vision crash course PART 1: Working with images in Python

In this 5 part video series you learn how to do machine learning for image classification within one hour. We recommend to spend a day or two to experiment with this before attending our other educations on conceptual machine learning and MLOps.

Video 1: Working with images in Python
Video 2: Setting up the Colab and GitHub environment
Video 3: Exploring the sklearn dataset
Video 4: Training and evaluating a classification model
Video 5: Predicting our own images and saving the model

The exercise uses Python, Google Colab and sklearn.

[sruizvargasp]

Hi Daniel. I'm Sergio Ruiz. First of all, congratulations for the contents. In mid 2010 I left it because I couldn't find anything about ACAP... and now, since yesterday I discovered you: it is an explosion of information compared to before.
I just wanted to ask you, because I've seen the Google Colab environment. I had to use it to train and develop the models. Once I have the model, I imported it into a raspberry.

My question is, are you running the models already trained on the AXIS cameras? I understand that on these cameras, their operating system will have some libraries tensorflow, keras, numpy.... built and compiled according to their architecture. I don't know, I just started with your videos, etc., but I already have 2 things in mind. Test the IA model directly on the camera (we will have to see if it supports, etc.)
And the second thing, I have seen that on the camera and on PLCs in Step7, the MQTT module or library works. Its communication would be very interesting, which until now was with the use of digital inputs/outputs that the camera had.

Surely you have already explained and developed the topic of installing third-party libraries, their compilation, etc. I will go video by video, consolidating the information. Not like before, if hardly any information

[davidwollover]

720 resolution + code font too tiny = difficult to read.