AI Project Cycle

AI project cycle is the life cycle of an AI project defining some steps that every organization should follow to develop an effective and powerful AI Project. There are five steps in AI project cycle as discussed below:

  1. Problem Scoping
  2. Data Acquisition
  3. Data Exploration
  4. Modelling
  5. Evaluation
AI project cycle on

Let us understand these steps with the help of example:
Imagine the situation!

The world’s largest “diamond” is in danger as Mr. Chor has threatened to steal it. No one is able to track Mr. Chor and so the situation is critical. You have been appointed as the Chief Security Officer to enhance the security of the diamond to make the area impossible for Mr. Chor to break into and steal the diamond. Now, prepare a system/project to accomplishing your task with the help of AI.
Here, you will find the problem, cause of problem, nature of problem, an idea to solve the problem.

For this example, we decided to make a facial recognition device that helps to identify the person.

The process of finalising the aim of a system or project means you scope the problem that you wish to solve with the help of your project. This is "Problem Scoping".

As you already decided the aim of your project. Now, you need to collect some informations required for your project. Needs to interact with the authorities to know about the people who are allowed to enter the area where the diamond is kept. Some of them being - the maintenance people; officials; VIPs, etc. Now, your challenge is to make sure that no unauthorised person enters the premises. Such information includes photographs, videos, personal details etc.

The process of collecting or acquiring the data in a visual form is known as "Data Acquisition". This data now becomes the base of your security system. Note that the data needs to be accurate and reliable as it ensures the efficiency of your system.

After acquiring the required data, you realise that it is not uniform, as it is difficult to analyize the large amount of un-structured data. Thus, you think of putting all the information collected in a simplified format and sequence. It means it is necessary to explore and arrange the required information from the large collected data in any specified format.

The process of interpreting or exploring some useful informations out of the large accquired data for a better understanding is known as "Data Exploration".

After exploring the data, now its time to develop a system or project which detects the face of a person and to match it with the existing image data you have in your system.
For this, you put all your explored data into the AI-enabled model that have different AI-enabled algorithms and train it in such a way that it alerts, if an unknown person tries to enter the vault.

The process of selecting and implementing the model which match the requirements of a project is known as the "Modelling".

After successfully implementing the model, Your surveillance system is now complete. Before deploying the project in the real-world, You must test it by sending a mix of known and unknown faces to the vault. After evaluating this model, you work on other shortlisted AI algorithms and work on them.

The final stage of testing the models is known as "Evaluation". In this stage, we evaluate each and every model tried and choose the model which gives the most efficient and reliable results.

After proper testing, you deploy your surveillance system in the premises. Mr. Chor, who is unaware of the surveillance system, tries to break through the vault and gets caught in your system. You have saved the diamond!