Project Card
Project
Build, train, test and deploy an Artificial Intelligence (AI) model to detect cats and chickens.
Value Proposition
A domestic animal identification system is essential in our university to meet the increasing demand for effective and dependable animal identification and monitoring. By employing machine learning algorithms, this system distinguishes between cats and dogs, offering a flexible solution for our research facilities, domestic animal owners, shelters, veterinary clinics, and wildlife monitoring initiatives.
Data
Inputs: Photos of cats and chickens
Output: domestic animals detection status
Endpoint Deployment Link
https://teachablemachine.withgoogle.com/models/Zrfl3TIb7/
Reflection
The objective to differentiate between cats and chicken was clearly defined and achieved?
In collecting the data, I browsed several webpages through Google Image and Yahoo image search engines to collect several and diverse species of domestic cats and chickens to train the teachable machine. There are a good number of images available online, of which 12 images of each of cat and chicken, were collected and used to train the system. These cats and chickens are of various colors and sizes. This was done to arrive at balanced results.
There are some challenges I faced during the exercise. For example, when either cat or chicken was used to test the machine, the result was correct with 100% accuracy. But when other animals different from cat and chicken are used to test the system, the system lowered the score to something below 85%. It will be better for the system to assign a zero (0) score instead of a lower score for images that did not match the input at all.
The system was impressive, indicating that AI can help one to produce learning machine system without the knowledge of coding language.