It's time to tell the story of the Artificial Intelligence Conference.
During the first two days of the conference, workshops and tutorials were conducted, where authors of their works presented new solutions in the field of artificial intelligence in computer vision. In my opinion, the most interesting workshop during the event was "AI for content creation", while the most interesting workshop was "Human-Centered AI for Computer Vision".
During the content creation workshop, the work mostly focused on showing the latest technology in the creation of avatars, which is currently very well developed by Apple. Also demonstrated were the capabilities of the GAN network for creating photorealistic images such as buildings or landscapes.
In the case of the tutorial, the lecturers presented in a very good way the problems currently encountered by us related to the interpretability or interact ness of the models created under humans. It was very informative in relation to my use of convolutional networks to recognize non-normative states in
elderly people.
The oral sessions, by design, were a very interesting idea to introduce the subject matter of many of the papers, which were then presented as posters. However, due to the number of performances and the broad subject matter, it was very hard to focus on the content being conveyed. One pleasant surprise was the work of MIT student Joanna Materzynska on word recognition using measuring the probability between natural text and the images presented images.
The most interesting element during the conference. During the poster sessions, not only could you quietly read everything about a given
work and download its code, but also talk to the authors of the papers. My attention was mainly drawn to the topics regarding pose estimation, in connection with the reference to my engineering work. It was a very interesting fact that the networks for human motion recognition are currently being taught much differently. We should also mention the PoseTrack21 database, thanks to which makes it much easier to train a model.
During the conference also came exhibitors from all over the world. It was a fantastic experience to see the most advanced
technology live. The most impressive were the cars autonomous cars and an application from Apple for creating visualizations of living spaces.
During the CVPR conference, I also had the opportunity to talk to people representing companies such as Snapchat, Pinterest, Apple, Amazon, and Disney. It was a very interesting experience because after a short conversation I would receive an offer to participate in qualifications for the company.
Going to the CVPR conference was an amazing experience, I did not expect that I would learn so many new things related to artificial networks, as well as the fact that there has been such a rapid development of the methodology in creating models for the human body pose estimation. I am very happy that I was able to participate in the workshops and tutorials, as well as talk to the developers of the projects because it made it I was able to make adjustments to the thesis in a much easier way. corrections to the thesis.
