Microsoft Hackathon 2021 Partner Group - 1st place
AI Recognition In Chicken Slaughterhouse
AI recognition applications that makes the slaughtering process humane for chickens and reduce production harm.
Overview
This is the project that won first place in Microsoft Hackathon 2021 in WiAdvnce Technology. In the project, I helped the chicken slaughtering factory introduce AI and human factors engineering to increase efficiency, reduce production injuries, and make the slaughtering process more humane.
Role & Responsibility
- Solo UI/UX Designer
- User Research (Interviews)
- UX Design (FigJam)
- UI Design (Figma)
- Prototyping (Figma)
- Presenter (Video, Adobe Premiere)
Problem
Stakeholder Interviews
In Taiwan, Chicken is one of the most popular food. Taiwanese can eat about 500 million chickens a year, and this number is still growing. As I interviewed potential users, namely managers and workers, in the chicken slaughterhouse, I found they faced the following problems during digital transformation:
With no data support, the supervisor does not know how to effectively select suppliers.
Supervisors need to monitor the efficiency and safety of workers in the factory, leaving them no time to make more meaningful decisions.
Workers are required to record the quality status of the chickens at a fixed location, which wastes manpower, and makes the data more likely to be wrong.
The food safety law is stricter than before, but the factory is still unable to build consumer trust through traceability of chicken food.
Create
First, I list all the steps in the factory. Next, I list the steps in which the problem occurred in order and discuss with the team how to resolve it.
How to choose good quality farms?
In the beginning, we asked managers what data they could use as a basis for selecting farms. Then, we use artificial intelligence to collect data in the production line, including the actual number of chickens, the defect rate, and the average weight. Actual numbers differing from supplier data, high rates of breast bruising defects, and average weights that are too light are all evidence that a farm is not up to standard.
How to keep workers clean and safe while hanging chickens?
After chickens arrive in the slaughterhouse, workers hang chickens onto a conveyor by hand.
In order to ensure that workers are properly cleaned and hang chickens with the correct SOP and body movement, we use human factors engineering technology to dismantle standard cleaning and hanging movement. When workers do not wash their hands properly, or the body movement is dangerous, the alarm will give a warning sound.
How to record quantity and defect data?
We used the dynamic image recognition algorithm for training and testing, and the recognition results were summarized to track down the quantity and defects of chicken.
Different defects have different causes.
Empty Hook: The workers did not hang the chickens orderly, which leads to an empty hook affecting the production capacity.
Head defects: When the bleed-out cut isn’t accurate, the chicken head may be cut off.
Wing defects: If the workers grab the chicken in an incorrect way, chicken wings may break or be bruised from rough handling.
Bruised breasts: This is caused by mistreatment of chicken in the farm, the data can be used as a reference for farm evaluations.
How to build consumer trust through traceability?
After the introduction of AI and human factors engineering, many traceability information data that have not been recorded in the past will be stored in the database which interact with manufacturing management systems, so in the event of an incident all affected products can be found quickly along the supply chain. These traceable data can be Printed on the packaging, helping to build consumers’ confidence in the safety of their food and drink.
Result
Improve supplier management
The data we collected on supplier dashboard plays as a key indicator for setting standards with scientific methods, and can improve supplier management.
Maintain cleanliness and safety
Use human factors engineering technology to track workers’ hand washing and chicken hanging actions to ensure the cleanliness and safety of the factory.
Save labor costs and track critical data
By using AI recognition to identify and record the quantity and quality of chickens, we reduced 264 working hours, which is about 1.5 manpower, per month.
Increase traceability of chicken food
Through the data collected during processing, the traceability of chicken food can provide customers peace of mind.