The workshop embarked on its journey into the topics of Machine Learning and Computer Vision at 11 AM, promising a captivating exploration full of practical insights and interactive experiences with Google Colab notebooks that were shared with the participants. The prerequisites for taking part in the workshop were that participants had curiosity and interest in the presented topics, but also knowledge of the programming language Python.
The journey kicked off with an electrifying dive into the topic of image analysis through cosine similarity. To learn this, the participants delved into the interesting world of pretrained neural networks and explored image pair comparisons with a dataset of apple-, car- and orange-pictures.
In the next step attendees learned how to download Images to Google Drive and create Embeddings: This session detailed the process of downloading paintings from the web, converting them into numerical representations, and delving into similarity searches and clustering techniques.
After successfully crafting the Image Embeddings, it was time for the next step: Image Similarity and Clustering of Street Artwork. This notebook focused on powerful techniques like image similarity metrics and clustering. Showcasing their utility in image analysis and data cleaning tasks.
But the adventure didn’t stop there! In the next step, participants explored the visualization of embedding space using TensorBoard, this session provided insights into understanding datasets and the embedding creation method.
In the grand finale of the tutorial, the attendees got introduced to running a Pretrained Resnet Convolutional Neural Network on Artwork Images.