Assessing the Mass Transfer Coefficient in Jet Bioreactors with Classical Computer Vision Methods and Neural Networks Algorithms

Development of energy-efficient and high-performance bioreactors requires progress in methods for assessing the key parameters biosynthesis process. With a wide variety approaches determining phase contact area gas–liquid flows, question obtaining its accurate quantitative estimation remains open. Particularly challenging are issues getting information about mass transfer coefficients instantly, as well development predictive capabilities implementation effective flow control continuous fermentation both on laboratory industrial scales. Motivated by opportunity to explore possibility applying classical non-classical computer vision results high-precision video records bubble flows obtained during experiment bioreactor vessel, we number presented paper. Characteristics bioreactor’s were estimated first (CCV) including an elliptic regression approach single boundaries selection clustering, image transformation through set filters developing algorithm separation overlapping bubbles. The application developed method entire filming makes it possible obtain parameter distributions dropout thresholds order better estimates due averaging. CCV methodology was also tested verified collected labeled manual dataset. An onwards deep neural network (NN) applied, instance segmentation task, has demonstrated certain advantages terms high resolution, while one tends be more speedy. Thus, current manuscript disadvantages discussed based evaluation bubbles’ their defined. coefficient virtue is represented.