Custom Machine Vision Software

Custom Machine Vision Software

ScienceSoft delivers custom image analysis software for machine vision systems, designed to meet complex needs of the manufacturing industry.

We partner with machine vision technology vendors, hardware manufacturers, system integrators, and end customers in manufacturing industries:

  • automotive
  • machinery
  • electronics
  • semiconductors
  • displays and LEDs
  • pharmaceutical
  • metals
  • construction materials
  • food, and more

Rooted in science, our software engineering competencies cover the entire process of image analysis in industrial settings: from image acquisition to communicating the final decision, we teach machines to see, understand, remember and, ultimately, learn. Our software applies to both PC-based machine vision systems and smart cameras, and can be developed as an embedded image analysis module for integration into third-party hardware.  

When specific manufacturing tasks require high-level customization, we step in to provide tailored solutions, recognizing our customers’ individual needs and a multitude of factors that influence them.

Get a quote

Image Analysis Applications in Manufacturing

With our industrial image analysis solutions, we enable every aspect of automated visual inspection (AVI) of both still and moving products and their parts.

Covering the inspection of both work-in-progress and finished products, we assist in quality control throughout the supply chain for better product traceability. We equip AVI systems with visual capabilities that no human eye can provide:

Object recognition

object recognition, classification and sorting

Object identification

object identification

Detection of defects and anomalies

detection of visual defects and anomalies

Assembly verification

assembly verification

High-precision measurement

high-precision measurement for dimensional accuracy

Counting in piles and stacks

counting in piles, stacks, and more (including touching and overlapping objects)

Barcode / label verification

barcode / label verification and validation, using optical character recognition (OCR) and 1D / 2D symbol decoding

Recognition of item location

recognition of item location and positioning

3D modeling

3D scanning, modeling and mapping

Reaching Business Goals

Enabled with powerful image analysis capabilities, machine vision systems bring accuracy, consistency and higher efficiency. Compared to human inspection, industrial image analysis solutions operate tirelessly, faster and more productive. As a result, it eases up human labor and lowers labor costs, increases yield, and with it, profitability. 

Our image analysis solutions help to:

  • Detect assembly defects and increase the output quality
  • Improve customer safety by preventing faulty assembly and mislabeling
  • Reduce workforce’s health risk resulting from manual handling of inspected objects
  • Reduce waste in the final product by eliminating assembly line and inspection errors
  • Shorten delivery time by speeding up inspection at all production and post-production stages
  • Comply with regulations, including on barcode grading, global traceability and mass serialization

Our Image Analysis Capabilities

We deliver custom software that supports every stage of the computer vision cycle for 2D and 3D images. We apply skill and talent to ensure precision and accuracy to minimize costly false rejects in any conditions, including high-speed motion on a conveyor belt. Our capabilities extend also to data coordination from multiple industrial cameras and simultaneous inspection of an object’s multiple properties.

Our solutions cover

Image acquisition

Image acquisition

Image restoration

Preprocessing: image restoration, noise reduction, deconvolution, enhancement

Feature extraction

Feature extraction

High-level processing

High-level processing of detected features to serve a particular application or a combination of applications at a time (e.g., in PCB inspection)

Decision making

Decision making, such in pass/fail and match/no-match tests


Communication of the decision from the machine vision system to the automation system in control via digital I/O, serial buses, protocols, and networks

Image sources

For image acquisition, we work with versatile image sources including the option to combine several sources in a single image analysis:

  • visible light, including LEDs
  • line scan imaging
  • infrared lights
  • microscopy
  • X-ray

Image processing methods 

To cover the variety of industrial machine vision applications, we offer the following methods performing high-level image processing in our solutions:

  • segmentation
  • pose estimation (geometric and learning-based)
  • object-based image analysis (OBIA)
  • image understanding
  • registration
  • pattern matching for location and verification using reference objects
  • quantification

Challenges We Solve

We recognize that for many manufacturers who are forced to adopt machine vision systems out of financial, ethical or regulatory considerations, there are still many unknowns when it comes to how reliable – and manageable – such systems are. We do our best to address the challenges and elevate industrial image analysis above misconceptions.

Low image quality

We consider lighting conditions and their limitations to incorporate required image enhancement at the preprocessing stage and achieve the clearest result possible.

Inaccurate analysis

With every analyzed image, we aim for a 99% and higher accuracy rate to minimize false rejects and overlooked errors on the production line.

Poor manageability

Our solutions can be managed by manufacturing floor staff with no specialized knowledge in machine vision thanks to user-friendly interfaces and flexible, configurable features.

Development Process & Technologies

For 30 years now, we have delivered custom software during 1,850+ projects worldwide.  Yet since its very foundation day, ScienceSoft has been dedicated to strengthening the grasp of C++ development as one of our core technological offerings.

Today, we employ 25 C++ developers with the average working experience of 11 years. To aid them, there is a strong team of software consultants who can take any rough idea up to its realization in a fully functioning solution.

Technological stack

  • C++ on Linux, Windows, OS X
  • OpenCV
  • tesseract-OCR
  • MatConvNet
  • R


  • Convolutional neural networks
  • Machine learning
  • Support Vector Machine (SVM)
  • K-means
  • ORB
  • Scale-invariant feature transform (SIFT)
Contact Us
Our image analysis software consultants are here to answer your questions, hold a free kick-off meeting and deliver a prototype to illustrate our approach and competence. Reach out to partner with us and stand out as a consistently reliable manufacturer.