Cashew Grading Automation in India

Challenges A cashew producer in India wanted to use image processing to automatically detect cashew grades based on size and remove low quality cashews.The company would need new hardware with a camera interface and a unique algorithm.The hardware would need to process 10 cashews/second, bringing a single cashew into a camera frame at a time. Xpertnest’s Solutions Designed a conveyer belt to deliver individual cashews to the camera, making processing and algorithm implementation easier and faster.Designed and developed an FPGA-based board to implement the algorithm and enable the high frame rate to process 10 cashews/second.Enabled continuous improvement with an optimised algorithm and...

Waterflow Control Using Remote Temperature and Moisture Sensors for a US-Based Agriculture Client

Challenges A US-based client wanted an Android app to collect farm data and automatically manage a water supply.Hardware needed to communicate with remote temperature and moisture sensors.An algorithm needed to automatically detect and analyse environmental temperature and moisture data. Xpertnest’s Solutions Xpertnest created a central hardware system that communicated with four points: RF sensors; pumps; an app; and a cloud server.A system allowed RF sensors to inform the central hardware but also communicated directly with RF sensing pumps.A custom-built app enabled universal monitoring and data access. Value Delivered Increase crop productivity and raised revenue.Reduced water and electricity usage, increasing conservation, lowering costs, and creating...

Survey and Analysis of Satellite and Drone Images for a US-Based Client

Challenges A US company wanted to find features on satellite and drone images.The company asked Xpertnest to design software to run an algorithm that could automatically stitch multiple images on a grid. Xpertnest’s Solutions Xpertnest created software to find correlations between adjacent images, and match and blend features.Adjusted the blended images for exposure and border smoothening.Made the stitched images available for deep learning algorithms in post processing. Capture images from satellite or droneEnter all images into a grid formatMatch features of adjacent imagesApply image blending algorithmsStitch individual images together and send them to Deep Learning algorithmsSurvey and Analyse Satellite Images Using an Image...

Automated Fruit Blemish Detection and Sorting Automation for a Spanish Agricultural Firm

Challenges A Spanish agricultural company wanted to be able to automatically detect grades for oranges using image processing to identify colour.The company needed unique hardware using a camera interface logic to run a sorting algorithm.The system needed to bring individual oranges into the frames of multiple cameras at the same time.Having graded each orange, the system would need to send the fruit to its respective container. Xpertnest’s Solutions Xpertnest developed a conveyer belt system to bring the oranges into frame.Designed and developed a processor to implement a sorting algorithm.The algorithm detected blemishes and differentiated between yellow and green skin.Enabled the possibility of...

Vehicle Number Plate Recognition for CCTV Vendor

Challenges A CCTV vendor wanted an algorithm that could recognize vehicle number plates from images, recorded video, and even live video streaming.The algorithm needed to be fast enough to detect numbers from fast-moving vehicles.The program needed to create an automatic entry and exit log for vehicles. Xpertnest’s Solutions Xpertnest developed a segmentation algorithm to localize number plates from vehicles detected from images, recorded video, and live video.Enhanced the number plate to apply thresholding and character segmentation.Applied an Optical Character Recognition(OCR) algorithm to recognize numbers on the number plate. Value Delivered The vendor could promote the system for maintaining vehicle records, detecting rules violations, generating...

Image Stitching for a United States-Based Company

Challenges A US company needed to stitch together microscopic images for deployment on an Android application.The images had uniform backgrounds and to some extent a uniform foreground too, requiring a robust algorithm to detect edges and stitch connections. Xpertnest’s Solutions Xpertnest developed an automated microscope that generated a grid of partially overlapping images.Created an algorithm that could match the features of neighbouring images whether rotated, scaled, or translated.Used the algorithm to stitch hundreds of overlapped image tiles to create a mosaic image Compute the translation of neighboring images Estimate image overlap and computer camera angle Replace computed translation with high uncertainty with estimated median value Optimize...

Text Detection and Recognition from Advertisements

Challenges A company needed to detect and recognize characters and numbers in the text of advertisements.To save server space, extracted information needed to be uploaded as text instead of as images.Detecting characters and numbers in different fonts and on a non-uniform background is particularly difficult. Xpertnest’s Solutions Developed a segmentation algorithm to localize text.Trained a machine learning model to detect text from images.Built an Android application to access a video stream from a server or capture an image, apply text detection and recognition, and upload text data to the server. Image processing of live video Identifying text from live video Recognize characters and understand meaning Store...

Real-Time, Multi-Object Detection for a Driverless Vehicle for a UScompany

Challenges A US-based company needed to identify in real time road hazards such as vehicles, pedestrians, strollers, traffic lights, and traffic signs.Identified objects needed to be fed to another system where a driverless vehicle could take action.To ensure accurate detection and classification, the system needed to be trained for multiple iterations and fine-tuned. Xpertnest’s Solutions Xpertnest trained deep learning models to detect and classify different objects.Used YOLO V2 architecture, Darknet19, and custom Network in Network (NiN) architecture for object detection and classification.Integrated a lane detection algorithm into the system. Value Delivered The YOLO (You Only Look Once) architecture enabled the algorithm to scan a...

Let’s Get in Touch

We’re interested in talking about your business.

A Company Registered in
England & Wales (CN – 10185772),
India (CN – U72100GJ2014PTC079101)
Ireland (Register Number – 686851)

Services

About us

Accreditations

© 2014-2022 Xpertnest. All Rights Reserved.