Case · Product · 2023 · Bangkit Academy
OPet
Too many pets are returned because the match was wrong. OPet pairs people with animals that fit their actual life, not just the cutest face.
- Role
- Machine Learning Engineer
- Stack
- CNN · Image Recognition · Maps API · Android
- Timeline
- 2023
- Team
- Bangkit C23-PS008
Too many adopted pets end up returned because the match didn't fit — OPet tries to fix that at the source.
The problem
People pick pets on impulse and by looks, then discover the animal doesn't suit their home or routine. The mismatch is bad for both sides.
What we built
A recommendation flow that matches adopters to pets by lifestyle, a nearby- adoptables map, and a CNN breed classifier that identifies a breed from a single photo.
Results
The breed recognition model reached 98% accuracy — built during the Bangkit Academy program alongside Google, Tokopedia, and Gojek mentors.
98%
Breed recognition accuracy