Summary:
Chemical and biological processes often involve the dissolution of substances in a solvent. This work explores an automated approach to monitor such dissolution processes, aimed at improving the overall efficiency of the process. A Faster Region-based Convolutional Neural Network (Faster R-CNN) with a ResNet50 backbone was trained to track and automated the dissolution of particles in various solutions. A dataset was created that incldes four different solutions, showcasing the dissolution process at various stages. The results show that the model can effectively detect medium to large particles, particularly those with minimal or no overlap with other particles. Additionally, the model was successful in monitoring colored solutions, such as potassium permanganate solution. The model's strenght lies in detecting medium to large particles. By expanding the dataset, detection accuracy for smaller particles can be improved, further enhancing the model's performance. Automating this process represents a critical step toward achieving fully autonomous laboratories.
for additional information please visit robovis.scitevent.org
