ML model for detecting contamination
#Industry #Python #Azure #OpenCV #PyTorch #MLmodel
Project definition
Location
Client
Project type
Industry
Service list
ML Engineering / Computer Vision
Team size
S-Size (1–3 engineers)
Budget
$50,000 – $250,000
Task
The goal of this project was to develop a machine learning model capable of detecting dirt in a reservoir and determining its quantity relative to the total volume. The model was trained using manually labeled data featuring reservoirs with varying amounts of dirt
Solution
The model was successfully implemented in production and deployed on the client’s server. The solution allowed for real-time monitoring of the reservoir condition, calculating the amount of dirt present, and determining its overall percentage. With an accuracy rate of 84%, the model enabled near-complete automation in tracking reservoir contamination, as well as the setup of an automatic alert system on the terminal for the addition of cleaning agents to the water
Impact
The deployed ML model automated reservoir monitoring with 84% accuracy, which reduced manual inspection efforts by over 33% and virtually eliminated human error in contamination detection. The system's automated alerts enabled immediate activation of cleaning protocols, minimizing production downtime and ensuring consistent regulatory compliance
💡 This is an AI-CORE project
💡 This is an AI-powered project
💡 Have a similar Industrial or Energy request?
💡 Have a similar Healthcare request?
💡 Have a similar EdTech request?
💡 Have a similar FinTech request?
💡 Have a similar request?
Detection of contamination levels
We developed an ML model to determine the amount of dirt in production tanks
Model accuracy is 84%
The high quality of the model significantly saves staff time and reduces the likelihood of unforeseen situations
Development Process
We designed and deployed a computer vision solution capable of analyzing real-time video streams from reservoir monitoring systems. The model, trained on manually annotated datasets, detects and quantifies dirt in water tanks under diverse environmental conditions. A full processing pipeline was implemented with Python, OpenCV, and PyTorch, supported by Azure and Apache tools to ensure scalability. The system outputs contamination levels with bounding boxes on live video, enabling operators to visualize results instantly and triggering automated alerts for timely action
Technologies
ML model for detecting contamination
Machine learning model for detecting contamination levels in tanks at manufacturing facilities
Book a call with our Head of Sales
