Ultralytics YOLO

YOLOvME: Zebra Species Detection

Explore innovative YOLOv5 applications, like zebra species detection, and see how our community leverages AI for challenging tasks.

ULUltralytics Team

2 min readMarch 21, 2022

A zebra detected by a YOLOv5 species detection model

We've seen people use YOLOv5 to create a crop yield estimation app, detect plastic in the ocean, and tell if someone is wearing their mask correctly. We reached out to our community and asked them to share even more ways of how they use YOLOv5 to solve their challenges.

Grevy's Zebra

Kayo Kumabe is a Data Analyst from Kumamoto, Japan. Kayo has only been working with YOLOv5 for the past month but enjoys experimenting with the infrastructure because “it’s just smart.” For someone who is new to AI, Kayo recommends that they spend time customizing their model, even if they don’t have knowledge of Python or machine learning. Generally, it can be difficult to distinguish subtleties in appearances with the human eye. Kayo hypothesized that instead, AI may easily be able to detect these subtle differences.

To test this, Kayo compiled a dataset of three different species of zebra to feed their YOLOv5 model. Kayo created a YOLOv5 model to detect different types of zebras. This model compares features of the animals and produces an output determining the species of the zebra.

As a result, Kayo was able to prove their hypothesis. The YOLOv5 model was able to detect each species of zebra with a high level of accuracy, while trained on just 20 images per species of zebra. We were curious to understand how Kayo got started with computer vision, so we asked them a few questions.

Mountain Zebra

Link to this sectionWhat Led You to Choose YOLOv5 for the Purpose of Zebra Species Detection?#

"I have never tried any other object detection infrastructure. YOLOv5 seemed easy since it did not require hard coding."

Link to this sectionWhat Aspects of YOLOv5 Made It Easy For You To Get Started With?#

"I collected only 20 images for each type of zebra, made labeling files of the image, and let YOLOv5 learn. That’s all! It is amazing to see YOLOv5 detected the zebra types 100% correctly! Maybe less than 20 images would be OK."

Link to this sectionHow Did You Get Your Start With Machine Learning and Vision AI?#

"Some of my clients are interested in AI so I started studying it to widen my job. When I came across vision AI, I was very excited because it looked like my favorite movie Terminator in real life."

Link to this sectionWhat Other Projects Do You Have In Mind For YOLOv5 In The Future?#

"I would like to detect my kid among many students in school. It could be useful on a sports day. I would like to make it for an iPhone application." To check out more of Kayo’s creations with YOLOv5, check out their LinkedIn and Twitter.

Link to this sectionHave Your Own YOLOv5 Use Case?#

This zebra detection YOLOv5 use case is a great example of YOLOv5’s success in distinguishing species. If we apply this neural network to other different types of animals, will YOLOv5 be able to differentiate those? How well will the model work if you want to detect pedestrians in a crosswalk or predict annual crop yield? Let your imagination run free! Tag us with #YOLOvME on our social media with your very own YOLOv5 use case and we’ll promote your work to the ML community.

Link to this sectionWant To Have Fun With AI Computer Vision But Want No-Code?#

All you need is an idea. With Ultralytics Platform, it's easy to create models with YOLOv5 and bring your ideas to life. We make things simple and do all of the complicated MLOps ourselves, so you don’t need to know any code to have fun AI. It's easy to get started and even easier to build your first ML model.