Track our process
Use machine learning to address the surplus of stray animals so that more animals in the shelter can find adopters, reducing euthanasia.
Team Members:
Xi Zhang, Siqi Jiang
Since we are unable to make any impact on the total number of stray animals in this project, we hope to use our knowledge to provide animal rescuers with a broader vision and thinking. We hope to help more stray animals in the shelter to find suitable adopters. Or by using deep learning models to predict the popularity of pets at the shelter, giving practitioners some direction, the adoption process can be made more efficient.
The goal of this project is to build a machine learning tool to predict how fast a pet is adopted so that shelters/adoption agencies can improve their pet profiles’ appeal, reduce animal suffering and euthanization, and better focus on their resources to help pets to find new homes.
The data was released by petfind.my in Kaggle on December 27, 2018. https://www.kaggle.com/c/petfinder-adoption-prediction/data
-
Feature Importance : XGBoost & LightGBM
-
Adoption Prediction: Turicreate - ResNet-50
- Implementation & Adjustment
- Mini Demo
-
Pet Statistics. (2020, February). Retrieved from https://www.aspca.org/animal-homelessness/shelter-intake-and-surrender/pet-statistics
-
Apple. (2020, April 30). apple/turicreate. Retrieved from https://github.com/apple/turicreate
-
Keras. (2017, December 12). ResNet-50. Retrieved from https://www.kaggle.com/keras/resnet50
-
PetMe uses AI to drive dog adoptions. (n.d.). Retrieved from https://www.ddb.com.au/petme-uses-ai-to-drive-dog-adoptions
-
Liao, S. (2019, February 11). Baidu made a smart cat shelter that uses AI to tell cats and dogs apart. Retrieved from https://www.theverge.com/2019/2/11/18220606/baidu-smart-cat-shelter-ai-facial-recognition-dogs
-
Build software better, together. (n.d.). Retrieved from https://github.com/topics/object-detection