tagmu icon

TagMu

‘TagMu’ assists financial institutions in livestock financing and insurance. It uses Computer Vision to identify animals and provide end-to-end verification for a fraud free livestock insurance. It assists both livestock owners and financial institutions through its one stop solution.
design
Design
development
Development
ios
iOS
android
Android
server side
Server Side
tagmu mobile app
For the algorithm design, we reviewed all the literature that is available over the internet for animal identification. We implemented various techniques and build multiple machine learning and deep learning models and analyzed their performance in terms of accuracy, robustness and efficiency. At last we selected the most accurate and robust of all our algorithms that was also efficient enough in response time to provide a good user experience when integrated with the application.
For mobile and web applications our process involved design thinking approaches including iterative prototyping and user centered design. We made prototypes of design and constantly included input of all type of users and stakeholders for the improving the design with each iteration. In the end we were able to achieve simple yet elegant interfaces for our applications that provide great user experience and seamless interaction.
tagmu cycle

Background

Livestock is bread and butter for most of the rural and semi-urban areas of Pakistan. Financial institutions provide livestock insurance and financing to farmers to secure their livestock businesses. However, the process being used for issuing insurance policies and verifying claims has been really hectic involving person involvement at different levels. It also requires physical presence of verification agents at the claim site that increases logistic costs and risks of fraud.
To address these issues and provide a hassle-free livestock financing solution to all stakeholders including Livestock Owners, Financial Institutions and Banking Sector, OneByte created a multi-platform application ‘TagMu’. It provides end to end verification of insurance claims through Computer Vision and allow all the stakeholders to manage the processes involved from Issuing Insurance Policy to last bit of verifying and approving claims.
tagmu mobile app
challenges

The Challenge Was To

One of the biggest challenges in automating the complete process of livestock insurance is to build a Computer Vision algorithm that is on one hand robust and accurate enough to minimize human intervention for claim verifications and on the other hand efficient and fast enough to provide good user experience.
Another challenge is the unavailability of relevant data and the effort required for collecting huge amount of relevant data and then curating it according to needs of the Computer Vision Algorithm.
Once that is done the next step is to build applications and interfaces around that algorithm, keeping in mind the diversity in background and technological knowledge of the stakeholders. On one hand we have banking and financial institutions that are somehow familiar with technological applications and on the other hand we have farmers and livestock owners that are mostly unfamiliar with the technological applications. Keeping in to account the needs and requirements of both groups and provide good user experience with a seamless design is another big challenge.

Features

  • Mobile application for Livestock Owners to view insurance policies, lodge claims and track claims.
  • Mobile application for financial agents to manage animals, insurance policies and claims
  • Claims verification using mobile camera through Computer Vision.
  • Geo Tagging of animals and agents to avoid fraud.
  • Web Portal for Financial Institution to track and manage agents, policies and claims.
  • Web Portal for Insurance Company to track and manage claims and policies.
  • Customized notifications for each application user based on its role and access.
  • Useful insights for users such as periodic stats for pending insurance claims, approvals etc.
features