Page 107 - AI for Good Innovate for Impact

AI for Good Innovate for Impact

enhancement is expected to improve the robustness and generalizability of the models across
different environments and conditions.

Second, the AI models will be further developed by integrating multimodal data, such as
combining gait analysis with vital signs or behavioral cues, enabling faster and more accurate 4.1-Healthcare
detection of health abnormalities. This approach not only aims to improve diagnostic precision
but also to extend the platform’s applicability to species beyond horses.

Third, a comprehensive animal welfare monitoring system will be established to support
continuous health tracking and disease prevention across multiple animal species [4].
Collaboration with veterinary institutions, farms, and animal welfare organizations will be
pursued to create a sustainable ecosystem in which AI technology contributes to proactive
animal care and protection.

Ultimately, through continuous innovation and strategic partnerships, AIforPet seeks to realize
its vision of enhancing the welfare of diverse animal species and promoting harmonious
coexistence between humans and animals.

3 Use Case Requirements

• REQ-01: The system should collect standardized images and videos to ensure consistent,
high-quality input for AI inference. The quality and accuracy of the input data directly
affects the performance of the horse identification and lameness detection models, so
all data should clearly capture relevant parts of the horse.
• REQ-02: The AI systems are meant to help and support veterinarians' clinical judgment,
not replace it. Diagnostic results should be used as reference information, and the final
decision should be left to the veterinarian.
• REQ-03: The system should be able to perform horse identification and lameness
detection simultaneously or separately, depending on user requirements. Operational
flexibility is essential to accommodate different use cases in a field environment.
• REQ-04: AI models must be continuously updated with newly collected data to improve
recognition accuracy for horse identification and early detection of lameness. Continuous
learning is critical to adapt to changes in breed, environment, and health conditions.
• REQ-05: The system should include mechanisms to encourage long-term monitoring by
notifying users when regular checks or new data collection is recommended. This feature
supports sustainable animal welfare practices by facilitating regular health assessments.

71