Recently ISSAI Data Scientist Askat Kuzdeuov was contacted by the popular Machine Learning (ML) and Data Science blog https://www.pyimagesearch.com to give them an interview. Askat has won a prize from them to become the first member of PyImageSearch University who completed all courses related to ML and computer vision.
The questions of the interview were mainly about the recent research projects that Askat is engaged in in ISSAI. In particular, Askat talked about the “SpeakingFaces: A Large-Scale Multimodal Dataset of Voice Commands with Visual and Thermal Video Streams publication, which is a dataset that computer vision practitioners can use to better understand and interpret voice commands through visual cue. In this project our researchers have initially realized that there was no available dataset which combined thermal (thermal camera), audio and visual data, so they’ve decided to design one themselves in order to expand the potential pool of applications of this dataset.
Askat also mentioned the obstacles they’ve faced with when collecting the acquiring the data for the dataset. For example, Askat talked about the ethics committee procedure, the tricky part of scheduling participant’s timings (as each participant had to spend a minimum of 30 mins of their time), and challenges with making sure that enough diversity is kept in participants’ pool, and how the pandemic has affected the data collection too.
In addition, Askat was asked about the use of GANs to translate thermal images to “normal” RGB representations of the faces, and pros and cons of using this method.
Askat was also asked about how he became interested in computer vision and deep learning, what are his recommendations for people who want to become successful researchers and specialize in data science too.
In his interview Askat also said that he would definitely recommend PyImageSearch University to any other people who are interested in learning computer vision, deep learning, OpenCV and other related topics in order to gain a knowledge of those concepts.
Please find Askat’s full interview here.