Intel AI Devcamp in Munich


Intel AI Devcamp

Intel AI Devcamp

Today was quite a rainy day, so I was lucky to be invited to the Intel AI Devcamp in Munich.
The reason I signed up for this event was to find out, what the current workflow of the AI tools is. Quite some time has passed since I last had some contact with this field. Six years ago I did a person detection based on SVM-classifiers and HOG-features embedded in OpenCV. Today’s workshop showcases Intel’s frameworks that are embedded in Python, Tensorflow and other APIs.

The devcamp started with the setup of the workshop attendees laptops, followed by some very brief introduction to machine learning and deep learning.
Before lunch, we did a first hands-on lab on an AWS-based Jupyter-notebook. The lab was named: Breeds Data Wrangling – Collect and Prepare Data and Setup for Training.

jupyter notebook

jupyter notebook


Our program was trained to look for a certain kind of animal breed.

After lunch, we worked in another hands-on lab in a VMware Ubuntu environment. Here we did a training of a CNN (Convolutional Neural Network) of the MNIST-dataset via CPU and the Movidius Neural Compute Stick. The advantage of the Movidius Stick is, that it basically performs nearly as good as a normal Intel CPU, but at a very low power (30W vs. 1W). Here our program was trained to look for the right numerical value of a handwritten number.

VMware Linux Tensorflow

VMware Linux Tensorflow

Movidius Neural Compute Stick

Movidius Neural Compute Stick

After the hands-on labs, two local startups showcased their products. The startup-company ProGlove impressed me most. They offer smart gloves, that are mostly used in car manufacturing. They have an integrated barcode scanner. This helps to reduce the time to document process steps during production.

Conclusion

It was quite an amazing day, lots to see and learn and eat 😉 The only issue for me was, that the learning curve is very high, so today’s event was not very beginner friendly. I will try to dive a little bit more into the Tensorflow environment in the next weeks.

Schreibe einen Kommentar

Deine E-Mail-Adresse wird nicht veröffentlicht. Erforderliche Felder sind mit * markiert.

Diese Website verwendet Akismet, um Spam zu reduzieren. Erfahre mehr darüber, wie deine Kommentardaten verarbeitet werden.