Entry No. 005
Junior (under 14)/ Senior (14 and above)
Senior (14 and above)
Category
6.2.1.1(6) Visionary Technologies
Classification
B. software
Country / Region Thailand
Name of Invention/Artwork Artificial Intelligence Systems For Livestock Counting on Farms
Outline of the entry
This project presents an AI-powered system designed to automatically count goats in a farm environment using image and video inputs. The system applies computer vision techniques, specifically OpenCV, for processing video frames, and YOLOv8, a state-of-the-art object detection model, to identify and count goats in real time.
The development process begins with collecting images and videos from real farms and online datasets. These visuals are then labeled using LabelImg, and formatted into YOLO-compatible annotations. The model is trained using this dataset to recognize goats under various lighting and environmental conditions.
Once trained, the model can be deployed to perform live detection via webcam or video files. The system outputs the number of goats on the screen and visually highlights each one with bounding boxes, allowing users to verify accuracy instantly.
The project addresses critical issues in traditional farming such as time-consuming manual counting and human error. By automating the process, the system helps farmers manage livestock more efficiently, with better accuracy and less effort.
The solution is scalable, adaptable to various animal types, and has the potential to integrate with smart farm systems in the future—for example, linking with feeding systems, health monitoring tools, or behavior tracking applications. This innovation showcases how AI can transform traditional agriculture into a more data-driven and intelligent practice.
Characteristics of the entry
In livestock farming, counting animals such as goats is a fundamental task that significantly affects management decisions, including feeding plans, health checks, and marketing. Traditionally, farmers manually count their livestock, which is time-consuming, labor-intensive, and prone to human error, especially in large herds or when animals move frequently.
With the advancement of artificial intelligence (AI) and computer vision, there is an opportunity to automate this process and improve accuracy. This project was developed to address the inefficiency of traditional livestock counting by creating an AI-powered system capable of detecting and counting goats in images and videos. Using OpenCV for image processing and YOLOv8 for real-time object detection, the system provides fast and reliable results.
The goal is to reduce farmers’ workload, minimize errors, and enable smarter farm management. The system can be applied not only to goats but also to other livestock types with proper data training. It offers great potential for scalability, integration into smart farming ecosystems, and future expansion into health monitoring and behavioral analysis. This innovation helps bridge the gap between traditional agriculture and modern digital solutions.
Demonstration details
1. Data Input:
Images or video clips of goats are collected from actual farm environments or online sources.
2. Frame Extraction:
Using OpenCV, the video files are split into individual frames for processing.
3. Data Labeling:
Each image is labeled using LabelImg software to identify the location of goats, and the data is formatted for YOLO training.
4. Model Training:
The labeled data is used to train the YOLOv8 model, which learns to detect and count goats in various conditions.
5. Real-time Detection and Counting:
The trained model is applied to new images or live camera feeds. The system processes each frame, detects the goats, and displays the count on-screen.
6. Output:
The number of goats detected is shown along with bounding boxes in real-time, helping users verify the system’s accuracy visually.
Other notes about the entry (if any)
This AI-based livestock counting system was developed as part of a student research project, with the aim to solve real-world agricultural challenges using accessible technology. The project demonstrates that even with limited resources, meaningful and scalable innovations can be created to support local farming communities.
One of the key strengths of the system is its flexibility. It can be adapted to count different types of animals—such as cows, chickens, or sheep—by retraining the model with new data. Moreover, the use of open-source tools (Python, OpenCV, YOLOv8, and LabelImg) makes the solution cost-effective and accessible to schools, universities, and small-scale farmers.
During testing, the system achieved an impressive average accuracy of 98.17% in goat detection. Its user-friendly interface and simple deployment method allow it to be used by people with limited technical knowledge.
Additionally, the project promotes interdisciplinary learning by integrating computer science, agriculture, and AI. It encourages students to apply technology to solve local problems and contributes to the growing movement of smart farming in Thailand and beyond.
This innovation is just the beginning. With further development, it can be integrated into broader farm management systems and scaled to benefit larger agricultural networks nationwide. แปล
Information on patent, utility model, trademark, etc. application
Number of Team Members
2
Student
Master Chassawee Billee May 23, 2007 34299@stw.ac.th
Master Korawit Sangaunprom June 12, 2008 34211@stw.ac.th