ISSUES The detection of blood cells is difficult due to the high image resolution, the variety of shapes of target objects, the large number per unit area, overlap, and size SOLUTIONThe YOLO-based detector was augmented with blocks to detect small overlapping objectsThe model, trained on a set of small number of images, yields results in predicting cells of different sizes and shapes, and is robust to noise caused by overlap, cell destruction and other anthropogenic factors MAIN FEATURES The model is capable of processing high-resolution images, detecting small objects, and is resistant to noiseAbility to process images with high separability and detection of small objects

The main goal of the project is to monitor and control the quality assurance of the contact centeremployees. The system solves issues such as the lack of progress of sales conversion, bias in theassessments of the internal team, penalties for non-compliance with regulatory standards byagents. Customized dashboards will allow managers to see the overall productivity of thedepartment, and thanks to the built-in ranking system, studying the effectiveness of a particularagent will become a matter of the moment. Using the unique technology of logging user responses, the process of analyzingthe interaction of the agent with the client becomes incredibly deep and reasonable, which will immediately spot and exclude problem areas.

ISSUES Lack of analogues on the market, the complexity of the problem due to the lack of a protocol with a classification algorithm, the lack of a unified methodology among specialists SOLUTIONUsing the existing SOTA network to estimate poses and simplify the architecture without loss of accuracy helps to run the program on non-resource intensive media  MAIN FEATURESA network built using a neural architecture search method that visualizes the importance of a feature in an imageReduced diagnostic time, improved diagnostic accuracy, convenience of using the application