Use our deep learning algorithms to detect quality defects like surface scracth, colour change, black dots, missing pattern etc. and deliver robust product to your customer.
Use our deep learning algorithms for pixel wise classification of object and work environment for robust and safe AI.
Enhance your safety and securty by deploying our deep learning algorithm to monitor safety practices, motion in prohobited areas etc.
Built an end to end robotics system using our vision system for your production processes.
Use our deep learning algorithm to understand consumer behaviour through their facial and body expressions.
Use Syntetic data to train your deep learning models and reduce the cost of data acquisition and project deployment. We partner you to create synthetic data for most challenging applications.
Girish has over 7 years of experience in the field of information technology and information security. He handled the production support and development teams for a major general insurance company based out of UK. He has also assisted several companies in Banking and Financial services sector to enhance their information security procceses.
The customer produces CTC wire as per their customer need. There are 3 variables that changes with each order and production ...
1. Pitch Length of CTC
2. Total CTC Length
3. Total No. of Transpositions
Challenge in Current Process:
1. Manual Error in CTC counts during continuous production.
2. High Production time due to manual stoppage while quality inspection to give breather to inspection person.
3. Pitch length measurement were not done completely and was done on sample basis which resulted in customer rejection.
4. Wastage of copper in case of rejection.
We discussed and analysed customer required and developed a vision system for quality inspection using deep learning algorithm. The developed solution enabled client to have 100% inspection instead of sample base inspection which resulted on only reaching high quality product to their customer. The algorithm did full inspection with 98% accuracy which resulting in substantial saving by eliminating copper wastage that was happening earlier due to defect detection at later stages.
One of the major Metal Recycler in India was facing a problem of accurate inventory accounting of their scraps. ...As a business process they used to receive mix scrap which was a combination 24-25 types of scrap.
The metal scrap is received as mix scrap which is sorted, stored and account as different types of scraps. The price of mix scrap depends on proportion of different scrap content it has. Higher component costs more else it cost less.
After sorting – classification was done manually by an expert.
The problem with the process:
1. Wrong classification by person specially during night shift leads to wrong value accounting of scrap.
2. Malicious attempt to classify inferior scrap to high quality scrap.
We took this challenging task and developed a computer vision system using deep learning algorithm to classify scrap based on their surface texture. The surface texture each time was changing but the building blocks of scrap remained same. The developed took around 8-9 month because of challenges of data collection but we were able to achieve 90% accuracy resulting in substantial cost saving for client.