Reduced Manual Effort
Automated reading of meter data decreased the need for human operators and reduced errors.
Technologies used
Python
Kubernetes
Tesseract OCR
TensorFlow
Keras
PyTorch
OpenCV
Pylibdmtx
Docker
Motivation
The customer faced inefficiency and human errors when manually reading data from meters, which include scanned graphs, barcodes, and handwritten notes. Azati’s goal was to automate the data extraction process, increase accuracy, and reduce the dependency on manual intervention.
Main challenges
Challenge 01
Round Disk Data Processing
The equipment printed data on round discs, which were scanned and sent to the system for reading and processing. Azati developed an algorithm to unfold the round disk and convert the image into a rectangular format to trace and read the coordinates of the curves accurately.
#1
Challenge 02
Curve Highlighting on Graphs
The graphs contained multiple curves with different colors, sometimes overlapping with the background. Azati trained a neural network to accurately select and highlight curves of different colors, even when extraneous interference occurred, ensuring accurate reading of each curve.
#2
Challenge 03
Data Format Variability
The customer had multiple partners using different types of equipment, generating data in various formats. Azati created a neural network that identified the input data's format, categorized it, and routed it to the correct data processor for processing.
#3
Challenge 04
Handwriting Recognition
Handwritten data, such as dates and numbers, presented challenges due to variability in legibility. Azati used Google Tesseract along with a trained neural network to recognize handwritten data from multiple regions of the scanned images, overcoming issues caused by human factors like haste or poor handwriting.
#4
Our approach
Prototype Development
The project began with the creation of a pilot prototype for reading curves on the graphs. After successfully achieving this, Azati expanded the system to recognize and process additional aspects of the input data, including barcodes and handwritten notes.
Coordination and Communication
Azati maintained continuous communication with the Canadian team for project management, prioritization, and delivery schedule coordination. This ensured that the project goals and timeline were aligned with the client’s expectations.
Iterative Development
The development process followed an iterative model, with data recognition services being built and tested in parallel according to the schedule agreed upon with the client. The team worked to refine the recognition accuracy with each iteration.
Facing the same challenge?
Bring your complexity. We'll bring the plan. Select a convenient slot to start a conversation with our experts.
Schedule a callSolution
01
High-Accuracy Barcode Processing
The system detects and reads barcodes from scanned images with over 90% accuracy. This ensures that critical information, such as equipment IDs and labels, is reliably extracted for processing and reporting.
Key capabilities:
- Automated barcode extraction from scanned images
- Recognition of multiple barcode types and formats
- Integration with downstream processing pipelines
- Error detection and correction for damaged or incomplete barcodes
02
Curve Recognition on Graphs
Curves on meter graphs are processed with over 80% accuracy. Neural networks identify each curve's coordinates and distinguish overlapping lines, allowing extraction of key performance metrics.
Key capabilities:
- Identify and separate multiple curves on a single graph
- Handle overlapping lines and noise in scanned images
- Extract coordinates for quantitative analysis
- Support various graph formats from multiple equipment sources
03
Handwriting Recognition
The solution recognizes handwritten numbers and dates from scanned images using Tesseract and custom neural networks. Accuracy depends on input quality, ranging from 30% to over 70%, significantly reducing manual data entry.
Key capabilities:
- Detect handwritten regions within images
- Transcribe dates, numbers, and notes
- Handle variations in handwriting style and legibility
- Support multi-region recognition for complex forms
04
Custom Neural Network Routing
The system identifies the input data format and automatically routes it to the appropriate processing pipeline, accommodating diverse equipment outputs from multiple partners.
Key capabilities:
- Detect and classify data format automatically
- Route data to the correct processing service
- Maintain high recognition accuracy across formats
- Adapt to new equipment and partners without system redesign
Results & business impact
Improved Accuracy
High-accuracy recognition of curves, barcodes, and handwriting increased reliability of operational data.
Operational Efficiency
Faster data processing improved decision-making and resource tracking for oil & gas operations.
Scalable Automation
The solution accommodates multiple equipment types and data formats, ensuring scalability across partners.
Foundation for Further Digitalization
Handwriting recognition improvements and pipeline automation set the stage for fully digital workflows.
