Automated Estrus Detection for Dairy Cattle through Neural Networks and Bounding Box Corner Analysis

Abstract

Thorough and precise estrus detection plays a crucial role in the fertility of dairy cows. Farmers commonly used direct visual monitoring in recognizing estrus signs which demands time and effort and causes misinterpretations. The primary sign of estrus is the standing heat, where the dairy cows stand to be mounted by other cows for a few seconds. Through the years, researchers developed various detection methods, yet most of these methods involve contact and invasive approaches that affect the estrus behaviors of cows. So, the proponents developed a non-invasive and non-contact estrus detection system using image processing to detect standing heat behaviors. Through the TensorFlow Object Detection API, the proponents trained two custom neural network models capable of visualizing bounding boxes of the predicted cow objects on image frames. The proponents also developed an object overlapping algorithm that utilizes the bounding box corners to detect estrus activities. Based on the conducted tests, an estrus event occurs when the centroids of the detected objects measure a distance of less than 360px and have two interior angles with another fixed point of less than 25° and greater than 65° for Y and X axes, respectively. If the conditions are met, the program will save the image frame and will declare an estrus activity. Otherwise, it will restart its estrus detection and counting. The system observed 17 cows, a carabao, and a bull through the cameras installed atop of a cowshed, and detects the estrus events with an efficiency of 50%.

Estrus Detection and Dairy Cow Identification with Cascade Deep Learning for Augmented Reality-Ready Livestock Farming

“…In another study, which was prepared as a doctoral thesis by Yildiz, an artificial neural network model was developed that used not only physical movements but also seasonal data, achieving 97% accuracy [ 27 ]. Arago et al aimed to detect cows in estrus that display mounting behavior, using models trained on images of cows [ 28 ]. Although the collected dataset has not been shared, a system that works with a 94% accuracy rate was developed with the help of the trained model.…”

“…If these symptoms are detected, only the success rate of artificial insemination increases [ 21 ]. While livestock wearables equipped with IMU, painting patches, and visual computing systems successfully identify mounting and/or standing-to-be-mounted behaviors, they are commonly referred to as estrus detection systems [ 28 , 29 , 30 , 31 , 32 , 33 ]. The proposed method is a visual system and is used to detect mounting and/or standing-to-be-mounted behaviors.…”

“…They trained single-layer artificial neural network models on this data and obtained an estrus period detection accuracy of 97% [ 27 ]. Arago et al developed a system that aims to detect cows’ estrus by tracking the mounting behavior displayed by cows during the estrus period [ 28 ]. In their study, they trained two artificial neural network models using the Tensor Flow Object Detection Application Programming Interface (API) with the goal of detecting the estrus event within 100 m. They carried out the detection process with the trained models by analyzing images taken from cameras installed at specific angles.…”

Practices and Applications of Convolutional Neural Network-Based Computer Vision Systems in Animal Farming: A Review

“…The situation is opposite for small batch sizes. Common batch sizes are in a power of 2, ranging from to in animal farming [ 91 , 120 ]. They depend on memory of devices (e.g., GPU) used for training and input image sizes.…”

“…Another influential factor is complexity of architectures. Arago et al [ 91 ] deployed batch sizes of 1 for faster R-CNN and 4 for SSD. Perhaps, complex architectures with small batch sizes can speed up training, while simple and lightweight large batch sizes can improve training efficiency.…”