Gomukh

Introduction

We have developed an intelligent mobile phone based biometric animal identification system ‘GoMukh’ that uses proprietary algorithms built on deep convolutional neural networks.

Animal identification is critical for proper verification of ownership, traceability of the animal produce, bio-security etc. A robust identification system is important for facilitating widespread adoption of livestock insurance policies that are needed to protect animal farmers. An unambiguous identity will greatly help establish trust between sellers and buyers creating a vibrant and fair market place to be benefit of the animal farmers.

Individual animal identification could be achieved either by mechanical, electronic, or biometric methods. The mechanical methods (e.g. using ear notching, ear tags, branding, and tattoos) are invasive method and not good enough for traceability purposes. Electronic-based methods mainly use external tags, Radio Frequency Identification (RFID) tags, to recognize animal but these tags are expensive and are subject to loss, removal, or damage. Biometric-based methods like iris scanning, retinal images or DNA analysis are also used for animal identification. Generally speaking biometric methods could give high identification rates, but they require specialized equipment and are not cost effective compared to other approaches like image-processing methods.

Biometric methods could give high identification rates, but they require specialized equipment and are not cost effective compared to other approaches like image-processing methods.

The mechanical methods (e.g. using ear notching, ear tags, branding, and tattoos) are invasive method and not good enough for traceability purposes. Electronic-based methods mainly use external tags, Radio Frequency Identification (RFID) tags, to recognize animal but these tags are expensive and are subject to loss, removal, or damage. Biometric-based methods like iris scanning, retinal images or DNA analysis are also used for animal identification. Generally speaking biometric methods could give high identification rates, but they require specialized equipment and are not cost effective compared to other approaches like image-processing methods.