Seeing is caring – automated assessment of resource use of broilers with computer vision techniques

Abstract

Routine monitoring of broiler chickens provides insights in the welfare status of a flock, helps to guarantee minimum defined levels of animal welfare and assists farmers in taking remedial measures at an early stage. Computer vision techniques offer exciting potential for routine and automated assessment of broiler welfare, providing an objective and biosecure alternative to the current more subjective and time-consuming methods. However, the current state-of-the-art computer vision solutions for assessing broiler welfare are not sufficient to allow the transition to fully automated monitoring in a commercial environment. Therefore, the aim of this study was to investigate the potential of computer vision algorithms for detection and resource use monitoring of broilers housed in both experimental and commercial settings, while also assessing the potential for scalability and resource-efficient implementation of such solutions. This study used a combination of detection and resource use monitoring methods, where broilers were first detected using Mask R-CNN and were then assigned to a specific resource zone using zone-based classifiers. Three detection models were proposed using different annotation datasets: model A with annotated broilers from a research facility, model B with annotated broilers from a commercial farm, and model A+B where annotations from both environments were combined. The algorithms developed for individual broiler detection performed well for both the research facility (model A, F1 score > 0.99) and commercial farm (model A+B, F1 score > 0.83) test data with an intersection over union of 0.75. The subsequent monitoring of resource use at the commercial farm using model A+B for broiler detection, also performed very well for the feeders, bale and perch (F1 score > 0.93), but not for the drinkers (F1 score = 0.28), which was likely caused by our evaluation method. Thus, the algorithms used in this study are a first step to measure resource use automatically in commercial application and allow detection of a large number of individual animals in a non-invasive manner. From location data of every frame, resource use can be calculated. Ultimately, the broiler detection and resource use monitoring might further be used to assess broiler welfare.

Review of Poultry Monitoring using Computer Vision

“…Mask R-CNN can be used to detect poultry and also make available pixel-level masks drawing their shapes. This can be beneficial in situations where comprehensive information about poultry contour is necessary (Eijk et al, 2022) (Chuang et al, 2021). Before choosing an object detection model for poultry detection, issues such as the available computational properties, required accuracy, real-time processing requests, and detailed characteristics of the poultry background should be considered.…”

“…In the realm of modern farming, technological revolutions are seriously improving the way we approach farming practices, trying to achieve more precision, sustainability, efficiency, and animal welfare (Neethirajan, 2022). The introduction of computer vision and image processing techniques into poultry management has led to a transformative force in industry (Zhuang & Zhang, 2019), (Eijk et al, 2022). This transformative solution has revolutionized the poultry industry by using innovative image recognition technologies to provide real-time insights into the health, behavior, and overall well-being of poultry populations (Eijk et al, 2022).…”

“…The introduction of computer vision and image processing techniques into poultry management has led to a transformative force in industry (Zhuang & Zhang, 2019), (Eijk et al, 2022). This transformative solution has revolutionized the poultry industry by using innovative image recognition technologies to provide real-time insights into the health, behavior, and overall well-being of poultry populations (Eijk et al, 2022). The genesis of poultry monitoring systems in the field of object detection dates back to the evolutionary period of computer vision and its introduction into various industries due to the need for non-invasive, accurate, and real-time monitoring systems (Zhuang et al, 2018), (Volkmann et al, 2022).…”

“…The technological architecture consists of coordination between hardware and software components to capture images with the help of cameras, process and analyze the visual data (Neethirajan, 2022), (Monitoring et al, 2004). Camera, which acts as the eye of the system, is connected in a stationary position above the poultry environment depending on the individual design and method adopted (Eijk et al, 2022), (Bao et al, 2021). The system captures activities of the birds within a defined environment, ranging from movement, interaction within the flock, water consumption, and feeding patterns (Rushen et al, 2012), (Massari et al, 2022).…”

“…Captured images are subjected to pre-processing steps, where noise is removed from the image, and the image is resized which will lead to enhancement of the image and extraction of all the key features, no matter how insignificant, such as nuanced behaviors and physiological features are recognized and interpreted for better outcome or result (Pereira et al, 2013), (Publishers, 2022). Implementation of deep learning models, which is the core of the system, is made possible through training of a vast dataset and validation of the model for better insights into the health (Neethirajan, 2022), (Eijk et al, 2022), (Fang et al, 2020). These will enable early detection of potential disease outbreaks and abnormalities for prompt and timely intervention (Zhuang et al, 2018), (Machuve et al, 2018).…”