A Supervised Deep Learning Model Was Developed to Classify Nelore Cattle (Bos indicus) with Heat Stress in the...

Area of Science:

• Agricultural Technology
• Animal Science
• Artificial Intelligence

Background:

• Real-time monitoring in agriculture enhances decision-making and reduces animal stress.
• Non-invasive technologies are crucial for monitoring livestock in varied climates.

Purpose of the Study:

• To develop a deep learning model for classifying Nelore cattle (Bos indicus) into thermal comfort and thermal stress groups.
• To assess the model's accuracy, recall, and specificity in identifying cattle under different thermal conditions.

Main Methods:

• Collected 676 samples from 30 Nelore cattle (18-20 months) between June and December 2023.
• Measured biotic variables: rectal temperature (RT) and respiratory rate (RR).
• Measured abiotic variables: air temperature (AT) and relative humidity (RH).
• Utilized a deep learning supervised neural network model for classification.

Main Results:

• The model achieved 72% accuracy and 72% recall.
• Specificity was low at 42%, indicating difficulty in identifying cattle in thermal comfort.
• Challenges may stem from class imbalance and insufficient input features for environmental adaptability.

Conclusions:

• Supervised learning models are valuable for precision livestock farming.
• Model performance can be improved by refining input characteristics and balancing data.
• Further research is needed to enhance the model's ability to differentiate thermal comfort accurately.