Smart Pig Farms: Integration and Application of Digital Technologies in Pig Production

Abstract

The prediction that the world population will reach almost 10 billion people by 2050 means an increase in pork production is required. Efforts to meet increased demand have made pig production one of the most technologically advanced branches of production and one which is growing continuously. Precision Livestock Production (PLF) is an increasingly widespread model in pig farming and describes a management system based on the continuous automatic monitoring and control of production, reproduction, animal health and welfare in real time, as well as the impact of animal husbandry on the environment. Today, a wide range of technologies is available, such as 2D and 3D cameras to assess body weight, behavior and activity, thermal imaging cameras to monitor body temperatures and determine estrus, microphones to monitor vocalizations, various measuring cells to monitor food intake, body weight and weight gain, and many others. By combining and applying the available technologies, it is possible to obtain a variety of data that allow livestock farmers to automatically monitor animals and improve pig health and welfare as well as environmental sustainability. Nevertheless, PLF systems need further research to improve the technologies and create cheap and affordable but accurate models to ensure progress in pig production.

1. Introduction

The demand for pork is constantly increasing due to the rapidly growing world population, and pork currently ranks second in terms of consumption [1]. The prediction that the world population will increase by more than 35% to almost 10 billion people by 2050 is contributing to this increase in production. According to the OECD [1], the demand for pork will increase significantly in the coming period, rising to up to 11 million tonnes by 2028, an impressive 9.3% increase compared to the current demand levels.

Due to the increasing demand for animal products, it is necessary to introduce new technologies and practices such as precision livestock farming (PLF) to ensure the sustainability of pork production [2]. Intensive pig production requires the continuous monitoring of pigs to collect information on their health status and production characteristics. The aim of implementing monitoring systems in production facilities is to ensure animal welfare, achieve better production results and increase economic profit. Smart farming systems are a concept that combines different technologies to improve and overcome the challenges of animal production. They combine the use of sensors, communication and tracking systems, robotics and artificial intelligence [3].

Smart farming systems and PLF are two interlinked concepts which use advanced technologies to optimize management in pig production and improve pig welfare and health. These concepts are based on advanced technologies such as the Internet of Things, artificial intelligence and advanced real-time data analytics. PLF and smart farming systems have different areas of application. PLF, for example, utilizes a wide range of technologies [4,5,6], while smart systems focus on specific areas and mainly use AI-driven monitoring systems [7,8,9].

PLF encompasses a range of real-time monitoring technologies that aim to control the temporal variability of the smallest manageable unit of production. PLF systems, therefore, aim for the fully automated continuous monitoring of animals through the use of new technologies [10]. Data from the farm can be collected using cameras, microphones, sound analysis systems or other sensors within the production unit [11,12] or radio frequency identification technology attached to the animal’s body [13,14]. Information collected about the animals on the farm and also a complete overview of the current status of the farm are stored in the operating system and are available to the farmer. By implementing the PLF system on farms, the continuous and automatic monitoring of animals is made possible. This provides producers and farmers with information that they can use to make decisions about feeding strategies to optimize production. Climate change is high on the list of the major challenges of the 21st century [15] and has a significant impact on the productivity of pig production systems [16]. According to Tullo et al. [17], pig production has an impact on water, soil and air quality as well as on the global climate and overall biodiversity. PLF systems are a measure that can be used to reduce the impact of pig production on the environment and ecosystem. The application of PLF allows for the continuous monitoring of the impact of pigs on the environment by groups or individuals with the possibility of the simultaneous adaptation of environmental factors to each segment of pig production. The main objective is to increase the production efficiency per animal through genetic selection and the precise breeding of animals to make pig production sustainable [18]. In addition, the introduction of the PLF system on farms leads to a reduction in emissions and a positive impact on the environment [19,20].

Although this review does not follow the strict framework of a systematic review, we adopted a modified Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) approach [21] to ensure transparency and rigor in the selection of the relevant literature. The systematic search for papers was focused on finding studies on precision livestock farming in pig production and the digitalization of the production process in pig production. The literature search was initially performed by a single researcher (KM) and was independently verified by a second researcher (KG). The literature search was conducted across the following databases: ScienceDirect, MDPI, ResearchGate and the Web of Science. The research period examined was 2011–2025, as we wanted to include as many publications as possible. The keywords used for the research were “pigs”, “precision livestock farming or PLF”, “RFID”, “IoT”, “applications”, “cameras”, “precision feeding”, “thermal imaging”, “digital twinning”, “electronic identification devices”, “sensors” and “computer vision system”. Boolean operators (AND/OR) were used to combine terms where appropriate.

Initially, 420 articles were identified from database searches. After removing the duplicates (60) and articles that did not comply with the inclusion criteria (200), 160 full-text articles were left. Of these, 46 articles were removed due to insufficient technical detail or a lack of relevance to farm-based production. Finally, 114 articles were included in this review. The process of selecting the articles is summarized in Figure 1.

The search terms were based on three categories:

• Species: pigs.
• Technology: precise livestock farming, IoT, sensors, digital twin, thermal imaging, cameras and computer vision.
• Application: monitoring, feeding, welfare assessment, automation and digitalization.

The inclusion criteria for the articles were peer-reviewed journal articles written in English, focused on domesticated pigs (Sus scrofa) and described or evaluated applicable on-farm digital/PLF technologies. The exclusion criteria were articles on species other than pigs without real-world validation and articles that did not address animal-level applications. The distribution of the included articles is presented in Figure 2.

This literature review provides an overview of digitalization and the impact of the available technologies for precision agriculture in optimizing pig production, with a focus on improving animal welfare, increasing production efficiency and reducing the environmental footprint. It also provides insight into the main benefits, limitations and prospects of implementing the PLF system in modern pig production.

2. Precision Livestock Farming: How Did It All Begin?

The technological development of PLF took place in the USA in the second half of the 20th century with the development of precision agriculture based on global positioning signals (GPS) [22]. The first mass application of PLF occurred with the development of individualized electronic milk meters for cows, which came onto the market in the 1970s and 1980s, years before the term PLF was even coined by Halachmi and Guarino [23]. According to Berckmans [24], PLF describes a system based on the continuous automatic monitoring and control of production, reproduction, animal health and welfare in real time, as well as the impact of livestock farming on the environment. The starting point of PLF is monitoring and analyzing biological responses, which are used to develop algorithms to control parameters in the production process [2]. When a problem occurs in the production process, a warning signal is activated so that the farmer can take appropriate action in the shortest possible time to solve the problem at an early stage [24]. To be as successful as possible, PLF requires ideal conditions for monitoring and controlling the processes. Collins and Smith [22] found that PLF technologies are mainly used in intensive production systems, while their application in extensive systems is less common, mainly due to the challenges of data collection in large and heterogeneous areas under highly variable conditions. Despite technological advances, traditional farming, where animals are monitored under human control, is still practiced in many countries [5,25]. Nowadays, there are a variety of technologies used to optimize production processes in the pig sector (

Table 1. Comparison of different precision livestock farming (PLF) technologies in pig farming.

Technology	Advantages	Limitations	Application	Future Improvements	Implementation Challenges
2D/3D Cameras	Accurate body weight estimation, behavioral monitoring, activity tracking	Sensitivity to environmental conditions (lighting, dust)	Large-scale pig farms for automated monitoring	Improved AI for better accuracy in poor lighting conditions	High initial setup cost, require technical expertise
Thermal Imaging	Non-invasive body temperature monitoring, estrus detection	Expensive, calibration needed for accuracy	Health monitoring and early disease detection	Cost reduction and integration with AI-based alerts	High cost of thermal cameras, requires regular maintenance
RFID-Based Feeders	Individualized feeding, reduced feed waste	High initial cost, infrastructure	Precision feeding strategies in intensive production system	Increased affordability, integration with cloud-based analytics	Require RFID infrastructure, potential signal interference
Sound Analysis Systems	Early detection of respiratory problems and stress	Advanced AI models needed for accurate interpretation	Large farms with high number of pigs	Enhanced deep learning models	Require extensive labeled sound datasets
Automated Waste Management	Reduces environmental footprint, optimizes nutrient recycling	Initial setup costs, requires suitable maintenance	Production farms aiming for sustainability	Smart waste processing units	Requires space and regulatory compliance
AI-Driven Predictive Models	Enable early intervention in health and nutrition management	Large datasets needed for accuracy, need for continuous refinement	Farms leveraging big data for predictive analytics	Expansion of real-time data collection and self-learning systems	Data privacy concerns, high computing power requirements
Digital Twin Technology	Simulates and predicts farm conditions in real time	Integration of multiple sensors and high computing power	Precision monitoring of farm environments	Enhanced AI-driven decision-making with real-time optimization	Computationally demanding, integration complexity
IoT-Based Smart Barns	Automate climate control, feeding and monitoring	High installation cost, network dependency	Large-scale automated pig production	Integration with blockchain for data security and transparency	Stable internet, cybersecurity risks

). These include cameras to determine body weight, behavior and activity [26], thermal imaging cameras to measure body temperatures [27], stations to monitor feed intake and body weight [28], microphones to track the vocalizations of pigs [29], sensors to detect lameness [30] and radio frequency identification for individual identification [31]. The application of the PLF system is designed to detect changes in the normal condition and behavior of pigs that occur as a biological response of the organism to an environmental stimulus.

Using RFID technology as a systematic tracking approach provides precise and automated information about individual animals. The collected data enable continuous monitoring without a direct line of sight, as is required by conventional barcode systems [32]. RFID tags enable the automatic recording of the feeding rhythms, water consumption, behavior and growth of pigs. In addition, the application of these systems can help in the detection of health problems, as deviations from the usual feeding patterns can be quickly recognized in order to intervene [33,34]. Challenges in implementing RFID systems include the significant initial setup costs, ongoing maintenance and troubleshooting requirements for the technology and potential issues with tag collisions if the system struggles to accurately read information from each tag simultaneously [35,36]. Computer vision is a promising tool for individual pig activity monitoring, but accurately tracking individuals using imaging remains a challenge. Visual detection, on the other hand, uses advanced imaging technologies to monitor animal behavior and interactions in real time and allows for rapid intervention when negative behaviors such as tail biting are detected [37]. The advantage of visual detection is that a larger number of pigs can be monitored simultaneously without the need to physically tag each individual animal, which reduces the cost of tagging [38]. A disadvantage of visual detection technology is the high initial investment in high-quality cameras and software for image processing and the possibility of it being influenced by environmental variables such as lighting conditions and the positioning of the cameras [7]. The integration of RFID technology and visual recognition will improve the individual tracking of pigs in large production systems. According to Yin et al. [39], visual recognition can improve the accuracy of RFID systems by confirming the identity of pigs and providing individual behavioral data. Their results showed that a combination of multiple devices achieved 97% accuracy in tracking individual animals. This system was developed to overcome the challenges of tracking pigs in a group system related to the possibility of losing ear tags or interactions that can cause multiple pigs to look similar. The results of their study show that the tracking of pigs worked 90 % of the time during the trial period. Also, the combination of RFID technology and algorithms for tracking and detecting pigs overcame the issue of multiple pigs having similar appearances and provided more accurate data for behavior and health management (Figure 3).

By analyzing the data collected over a period of time, it is possible to create an automatic classifier that identifies the patterns that lead to differences in behavior due to inappropriate conditions [7]. It is important to correctly decipher and assess the value of the data collected, as this allows the animals’ biological responses to be used as an indicator to obtain information about the environment and stimuli. PLF enables efficient and sustainable resource use by reducing waste and allows for the early detection of diseases based on changes in animal activity and behavior monitored by cameras and various sensors, protecting animals from environmental stressors and reducing human manipulation, which improves their welfare [25]. In precision agriculture, "digital twinning" has recently been explored. Digital twinning is defined as a virtual system model which combines various technologies, such as artificial intelligence, the Internet of Things (IoT), communication and embedded technology, data analytics, security and cloud computing, and is used to plan, monitor and improve strategies in individual agricultural production [40,41,42]. Such a system offers good potential for application in pig farming, as it could be used to predict positive and negative changes in production, as well as to assess the potential health risk due to variations in environmental factors. Advanced data analysis within the digital twin allows for a better understanding and better predictions of pig diseases, which can lead to timely interventions and a reduction in the disease incidence. This improves the overall animal welfare and reduces the need for antibiotics, which are an important aspect of modern pig farming [43]. The digital twin system enables the creation of detailed simulations and scenarios that give breeders an insight into the possible outcomes before changes are implemented in the real world [44].

3. Implementation of Technologies

3.1. Pig Identification

PLF technology can be identified as performing various functions in the pig production cycle, some of which are the identification of pigs, the measurement of body weight and body temperatures and the application of precise feeding. Information on the implementation of different PLF technologies and their specific goals and practical applications is presented in Table 2.

To improve the welfare of pigs and the production system, it is necessary to analyze the behavior and activities of pigs throughout the production process [24]. Pig identification has been carried out for centuries with the aim of detecting and controlling various infections and diseases. Some of the techniques, such as shearing and tattooing, that were used in the past to identify pigs are also still used today [45]. More recently, ear tags were used, and in recent years, microchipping and the use of electronic identification devices (EIDs) have been introduced [2,46]. The first use of cameras to identify pigs involved the use of 2D cameras capturing grayscale images, which was limited by the lack of depth information [47]. These 2D cameras were later replaced by 3D cameras, which could capture three-dimensional images with less influence from the environment and lighting [48,49]. The intensification of pig production and the introduction of PLF technologies on farms have led to the development of a fast, accurate and simple pig identification system. With the modernization of pig production, the identification of pigs has shifted to using electronic identification devices. A Radio Frequency Identifier (RFID) device is one of the most commonly used devices for individual identification. RFID technology works on the principle of transmission between the device and the label. It uses radio waves to transmit data between the device and the tag, enabling the fast and accurate identification of each individual animal [50,51]. One of the main disadvantages of RFID technology is its limited operational range. According to Hansen et al. [52], long-range RFID readers can be activated and successfully read at a maximum distance of 120 cm. When multiple RFID tags are used in a single pen, the data accuracy can be as high as 97%. However, the application of RFID tags is associated with significant disadvantages. One of these is the pain and stress that can be caused when attaching the tags to the pigs. In addition, the RFID tags can become damaged or lost over time, rendering them useless for the intended monitoring purposes [53].

In addition to an RFID system, other technologies are also used for the individual identification of animals. Optical character recognition uses cameras and advanced algorithms to identify individual pigs. High-resolution cameras and facial analysis software are used to identify pigs based on the unique characteristics of their faces [54]. More recently, deep learning methods for pig face recognition have utilized the skin and texture features of pig faces for individual identification [47]. Implementing this technology is quite challenging as pigs are in constant motion and facial recognition is further complicated by the dirt that often covers the animals. According to Marsot et al. [55], the accuracy of the identification of pigs with 2D cameras is around 90% and increases to 96.7% when facial recognition technology is used for identification. These results suggest that a facial recognition system could be used as a replacement for commercial ear tags. According to Kang and Oh [56], the rapid development of digital technologies is leading to the continuous monitoring and observation of animals without human intervention. This has led to increasing automation, which contributes to the progress of precision livestock farming. Automation in pig production and smart systems play different roles in improving management efficiency and animal welfare, but they are both integral functions of technology used in pig farming [57]. Automation simplifies and streamlines processes and reduces human labor through mechanization, while intelligent systems rely on advanced data processing and adaptive technologies. In their study, Cowton et al. [58] implemented a deep learning system to recognize and individually track pigs on a commercial farm. They combined a deep learning method for object localization, the Faster Region-Based Convolutional Neural Network (Faster R-CNN), with two potential real-time methods for tracking multiple objects to create a complete system capable of autonomously locating and tracking individual pigs and enabling behavior monitoring via RGB cameras. The study showed that deep learning algorithms enabled the development of a relatively low-cost system for tracking pigs. The system also provided additional information on factors such as the behavior of the animals, the total distance traveled, the average speed and the resting times of the individual pigs.

One of the most important health and welfare problems in intensive production systems is animal aggression, which McGlone [59] defines as a complex interactive behavior that can last from a few seconds to several minutes. It manifests itself in the form of pushing, hitting, tail and ear biting and cannibalism and often occurs when the living conditions in such systems are inadequate [60,61]. These behavioral changes are usually accompanied by vocalizations by the pig and can be monitored and tracked with a special device, the Pig Cough Monitor. The Pig Cough Monitor is used in combination with 2D and 3D cameras to capture environmental sounds and plays an important role in the prevention and early detection of changes on the farm [62]. According to Wei et al. [63], research into the aggressive behavior of pigs is still at an early stage, and it is believed that some forms of aggression could be more easily detected using deep learning models such as those in the YOLO (You Only Look Once) family, including YOLOv4, YOLOv5, YOLOv6, YOLOv7 and YOLOv8 [63]. To improve the accuracy of pig recognition, Shao et al. [64] used YOLOv5 and DeepLab v3+ to create the world’s first annotated dataset for pig posture recognition, covering four basic postures: standing, lying down, lying on their side and lying on their back. An experimental study by Li et al. [65] found that among these models, the YOLOv5-KCB algorithm had the highest accuracy in recognizing individual pigs and showed a significant improvement over the original YOLOv5 algorithm, with a 4.8% increase in the head and neck recognition accuracy and a 13.8% improvement in face recognition.

3.2. Body Weight Estimation

The estimation of pigs’ body weight in intensive production systems ensures accurate performance monitoring and welfare assessment [66]. The use of PLF to estimate pigs’ body weight enables the more accurate and efficient management of pig breeding and feeding. This method uses various sensors and algorithms to analyze images and data and enables automatic weight estimation without the individual weighing of the animals. Changes in the body weight of pigs are an indicator of their health status. Therefore, monitoring these is important to ensure the health of the pigs [67]. The impact of PLF on health monitoring is important from the perspective of early disease detection and prevention, improved biosecurity measures and reduced reliance on antimicrobials. Racewicz et al. [68] emphasized that PLF systems are practical from a scientific perspective for detecting health problems in pigs, but are not yet trusted from a farmer’s perspective. The problem of convincing pig farmers to adopt new technologies that are economically viable is still unresolved. With its real-time continuous monitoring capabilities, PLF promotes pig welfare and also increases the efficiency and profitability of pig farming. Sensors, cameras and analytical systems are used to continuously monitor behavioral parameters that can serve as early indicators of health problems. Alexy et al. [34] used RFID technology to monitor Mangalica breeding sows in a pastured farming system. The authors demonstrated that the health and welfare of the sows could be improved by tracking individual animals and that health problems could be responded to in real time. The feed intake and weight gain are linearly related. Pezzuolo et al. [69] point out that monitoring pigs’ weight is an important factor in optimizing feeding costs and detecting problems due to insufficient or excessive feeding. Wang et al. [70] found that pigs’ weight had previously been estimated using visual and tactile methods incorporating body measurements, which were inaccurate. The measurements were made by collecting data on specific parts of the pigs’ bodies, which were then converted into an estimated body weight using mathematical calculations. Later, body weight was measured through individual weighing, which provided much more accurate values. Determining body weight through individual weighing requires the presence of two workers and takes an average of five minutes per pig [71,72]. This confirms that this method is time-consuming and stressful for the animals, which can lead to weight loss and various health problems. Therefore, this method is now mainly used on small farms. Nowadays, there are several methods for measuring body measurements and predicting body weight, such as Weight-Detect, Pigwei, eYeScan, Growth Sensor and OptiSCAN [73]. The machine vision-based weighing of pigs has several advantages over individual weighing with a livestock scale. When using a livestock scale, the animals have to be moved, which leads to stress for the animals, manual labor and a possible malfunction of the equipment. Visual image analysis (VIA) is a technique based on capturing high-resolution images of pigs’ body surfaces and analyzing them using specific algorithms. Mluba et al. [74] stated that VIA provides information on weight gain, as well as about the pigs’ health and welfare. One of the advantages of using VIA is the capability for monitoring without disrupting the pigs’ natural behaviors.

One of the first uses of cameras to estimate body weight was recorded by Schofield [75], who imaged a pig from the side and from above using a mirror placed at a 45° angle perpendicular to the pen. He estimated the weight of a live pig with an error rate of ±6.2%. After this research, many researchers used CCD cameras to determine the body measurements and body weight of pigs. Later, the application of 3D cameras using a fully automated system started, with the aim of facilitating the monitoring of the live weight and growth [76,77]. Vranken and Berckmans [62] describe an automated method for determining the body weight of pigs using video image analysis with 2D or 3D cameras. The system works by localizing the pig’s body within the system, meaning the camera must distinguish the pig from its surroundings. Body characteristics such as the body surface area and length and width of the body are then measured and converted into the body weight using mathematical calculations with the help of software, with a possible deviation of up to 1.5 kg. Win et al. [78] stated that images taken from above provide information about the growth of animals, which is why the cameras are usually installed in the upper corners of facilities.

Fernandes et al. [79] developed an automated computer vision system (CVS) capable of extracting variables such as body dimensions and descriptors of body shape from 3D images. These features were tested as potential predictors of body weight in live pigs. The body dimensions of the volume, surface area and body length extracted from the images showed the highest correlation with the body weight, while the width and height were highly correlated. To validate the accuracy of body weight measurements performed with new technologies and evaluate their applicability on farms, Abner [80] conducted a study based on three measurement methods: human observation, a platform scale (CIMA; Correggio, Italy), and PigVision mounted cameras (Asimetrix Inc; Durham, NC, USA). The accuracy of the human estimate was 88.2% and the scale measurement was 98.2%, while PigVision achieved an accuracy of 96.6%. Although the scale measurement was the most accurate, PigVision was the most suitable measurement method as it continuously provided relatively accurate and reliable data.

3.3. Body Temperature

In recent decades, studies have focused on obtaining reliable body temperature data without significantly manipulating the animal. As a result, the rectal determination of the body temperature in pigs is used less and less, especially in intensive rearing systems. Zhang et al. [81] found that one of the most economical and effective methods is the use of infrared technology, which allows for the continuous monitoring of the thermal biological response or thermal changes in the animal’s body. It is a non-contact, non-invasive method based on the principle of detecting infrared radiation and converting it into electrical pulses. The results are thermal images that show the distribution of the body’s surface temperature [82,83]. Thermal imaging cameras are used in various areas of pig production, e.g., to detect disease and estrus [84], to monitor health and welfare [85] and to measure body temperatures without using a rectal thermometer [86]. The advantages of using thermal imaging cameras are the high temperature accuracy, stability and simplicity of the system. Rectal thermometry and thermometry with embedded temperature sensors are conventional methods for monitoring temperatures in pigs, but they have several disadvantages [86]. Their use increases the risk of disease transmission via the equipment or devices [84] and at the same time increases the physiological stress response, causing discomfort and compromising animal welfare. Thermal imaging camera models suitable for different price ranges and requirements can be used to measure the body temperature of pigs. These systems rely on high-quality InfiRay thermal imaging cameras that enable non-contact, precise and fast multi-point measurements. These cameras can be installed in robots for the continuous temperature monitoring of all animals. Another innovative instrument is iBut-tonne®, a stand-alone system for measuring and recording the temperature and humidity. It can be attached to an identification tag on each animal’s ear, establishing direct contact with the inner ear, which is sensitive to temperature fluctuations, ensuring high measurement accuracy [87,88,89]. There are also various implantable and wearable devices, such as subcutaneously applied microchips or sensors embedded in ear tags [90], which offer high precision with continuous data collection [91].

It is important to recognize the impact of climate change, which increases the exposure of animals to disease, heat stress and numerous health risks. One of the main causes of heat stress in pigs is their limited ability to thermoregulate. Extreme temperatures lead to their body overheating, which has numerous negative consequences for both the animals and the producers. The use of smart technologies to monitor the feed intake, body temperature, respiratory rate and animal mobility can help prevent heat stress and directly improve animal welfare and productivity [92]. Chung et al. [93] found that anatomical areas with less hair allow for a more accurate measurement of the skin temperature, while Štukelj et al. [94] confirmed that the ear canal, external ear and perianal area are equally suitable for temperature measurement by thermography. Monitoring fluctuations in the body temperature is of great importance for ensuring optimal reproductive performance, as it increases the success rate of artificial insemination [93]. According to Sharifuzzaman et al. [95], reproductive failure due to unrecognized estrus behavior is one of the main causes of sow culling. The reliable detection of estrus is made possible by tracking the body temperature, movement patterns and behavioral changes, including vocalizations and activity. This is supported by the use of accelerometers to monitor activity, infrared thermal imaging to measure the body and vulvar temperature and radio frequency identification. Studies using accelerometers have shown increased activity levels in sows and gilts during estrus [91,96]. Wang et al. [97] and Zhang et al. [81] found that infrared cameras detected a significant increase in the average and maximum skin temperature of the vulva during estrus compared to on other days of the reproductive cycle. When the tail of a pig is raised, the vulva remains clearly visible, which facilitates temperature monitoring with cameras. In addition, Chem et al. [83] conducted a study in which no statistically significant difference was found between the temperatures measured with a thermographic camera and those determined using direct thermometer readings. They concluded that the DITI camera could serve as a new technology for heat detection in gilts.

3.4. Precise Feeding

According to Gaillard et al. [98], the success of pig production is reflected in the efficient conversion of feed into growth, which is referred to as the feed conversion ratio (FCR). Factors such as the feed wastage, digestibility, growth rate, intake and feed utilization have a significant impact on the FCR. To improve the FCR, the use of smart technologies in all areas of pig production is essential. In precision pig feeding, feeding is adjusted based on individual needs and conditions, using advanced technologies to collect and analyze data. PLF and precision feeding are advances in pig farming that aim to optimize feeding practices and improve animal productivity, health and welfare [98]. Since feed accounts for up to 70 % of the total production costs in pig production, precision feeding is an important component of PLF [18]. It implies the use of precision feeding techniques that provide the animals with a diet adapted to their production goals. The application of precision feeding aims to utilize nutrients without compromising performance. In addition, precision feeding facilitates adaptation to the needs of individuals or small groups in real time [99]. To implement precision feeding on a production farm, it is essential to install measuring equipment to monitor the feed intake and pigs’ body weight [100], as well as calculation methods and a feeding system that enables an individualized approach and allows for the optimal amount of feed for each animal. Since feed quantity and composition requirements vary depending on the production category, precision feeding utilizes automated feeding systems based on the recognition of an animal’s electronic identification response as it approaches the feeding unit [101]. The transmitters are usually RFID devices. According to Arulmozhi et al. [25], RFID tags play a crucial role not only in pig identification but also in precision feeding. RFID tags are placed near feeding and watering equipment to collect data on the number of visits and their duration. When an animal approaches the feeder, the reader recognizes the identification tag and dispenses the appropriate amount of feed. The feeding time, the amount of feed consumed and the number of visits to the drinker are recorded [102,103]. A computer system collects and stores all these data, enabling the precise monitoring of individual feeding behavior. The system allows for individual access to the feeding station, and for larger groups of pigs, a larger number of feeding places are provided to ensure optimal feeding conditions. As setting up individual feeding and monitoring systems is not cost-effective, a reliable camera system is usually used for this purpose. Although the monitoring of behavioral patterns in pig feed and water intake has not yet been carried out on a large scale, all monitoring to date has been carried out using cameras, mostly 2D and 3D cameras. Chen et al. [54] have proposed a deep learning method based on Xception and LSTM networks to detect the feeding behavior of pigs. This method can differentiate feeding episodes from non-feeding episodes and determine the feeding time of each pig with a precision of 95.9% and an accuracy of 98.4%.

Gaillard et al. [98] emphasize that in commercial herds, individual identification is usually carried out for sows. According to Aparicio et al. [103], the reason for this is challenges in feeding management in farrowing barns. Their research found that the use of electronic sow feeders (ESFs) during lactation significantly impaired the production performance of both sows and piglets compared to conventional feeders. The study also showed that the sows lost body weight during lactation and had higher milk production and a higher lactose concentration in their milk. The results also showed a decrease in the feed consumption per kilogram of weaned piglets, which is of great economic importance. Andretta et al. [102] found that the precision feeding of fattening pigs had an impact on nitrogen and lysine excretion. The authors reported a 26 % reduction in the lysine intake and a 30 % reduction in nitrogen excretion. Precision feeding increases animal productivity, has a major positive effect on the environment and improves economic sustainability. In their study, Perez Garcia et al. [104] developed a prediction model for the indoor temperature on a farm with a 72-h forecast horizon to obtain information on temperature fluctuations within farms. This approach and the use of PLF made it possible to improve the overall efficiency and integrate the predictive capabilities of the model into the production process. Andretta et al. [105] investigated the impact of precision feeding systems on the environmental impact of Brazilian pig production. Daily precision feeding through group and individual feeding reduced eutrophication by up to 4 % and acidification by up to 3 % compared to a conventional feeding system. Individual feeding showed better results compared to conventional feeding and reduced the potential impact on the climate by up to 6% and the potential impact on eutrophication and acidification by up to 5%. Llorens et al. [106] also used an individualized precision feeding system to assess the impact of a precision feeding system on the environmental impact in the rearing and fattening phase of pig production. The authors found that individualized precision feeding reduced nitrogen excretion by 30% and phosphorous excretion by 40% and also reduced nitrous oxide and methane emissions from the pens and manure. Mielke and Stein [107] investigated the integration of convolutional neural networks (CNNs) and transformation models to detect excreta in pig pens. The authors showed that the application of AI-based computer vision methods opened up new possibilities and methods for measuring and predicting emissions from livestock farming and promoted the use of this technology in measuring, modeling and mitigating emissions. Papadopoulos et al. [92] stated that sophisticated data analysis, automated feeding systems and precise monitoring are essential to meet the objectives of the European Green Deal and its "Farm to Fork" approach. This strategy aims to reduce greenhouse gas emissions, improve animal welfare and promote sustainable farming practices by 2023.

4. Future Perspectives

Pig production is one of the most technologically advanced branches of production and is constantly growing. In recent years, the introduction of advanced technologies and management methods has led to an increase in the number of pigs and a simultaneous decrease in the number of breeders [2]. In recent decades, the number of small farms has decreased. One of the reasons for this trend is the consolidation of small farms. In addition, it is common for large producers to take over small farms, which leads to the greater automation of farms. Nowadays, minimizing the negative impact on the environment is the main goal of pig production. The integration of renewable energy sources and the recycling of materials and pollutants from pig production are taking place in intensive pig farming [108]. In order to maximize feed efficiency and reduce waste, using technologies such as automatic climate control systems and precision feeding can help to achieve these goals. As this production process is extremely labor-intensive, more and more farmers are turning to PLF systems to facilitate the monitoring and management of production [109]. The development and advancement of modern technology has opened up a whole spectrum of possibilities. Wireless data transmission is readily available, cheap and reliable. Moreover, the technologies required for the development of PLF systems are now small and reliable enough to be used in challenging livestock farming conditions [24,48]. Machine learning algorithms are increasingly being used to analyze real-time data and enable early disease detection and better herd management. Today’s computer systems can continuously monitor any parameter of interest of a pig or group of pigs inside or outside on the farm in real time [2]. Such automated systems are available 24 h a day, seven days a week. The use of smart agricultural technology in pig production therefore has significant potential to visualize the impact of various changes on pig production, health and welfare, as well as the impact of pigs on the environment.

Bordignon et al. [110] conclude that smart technologies such as sensors, data analysis and automated systems can also play a key role in the adaptation of pigs to climate change. Pigs are homoiothermic animals that need to maintain a relatively constant body temperature in a wide range of climatic conditions [111]. The consequences of climate change are therefore particularly evident in the occurrence of heat stress [112], which has a negative impact on the welfare of pigs and their production characteristics and causes major economic losses [113]. The efficiency of a farm can be improved by integrating digital IoT solutions, which also leads to better decision-making processes in relation to the production cycle. Therefore, the future of the PLF system in pig production will involve information and communication technology, sensors, cameras and microphone systems that can monitor and observe behavior and biological responses to various stimuli in the environment. These advances will not only improve animal welfare but also contribute to economic efficiency by reducing manual labor costs and optimizing resource allocation.

Table 2. Precision livestock farming (PLF) technologies applied in pig production and their application in production cycles.

Parameter	Applied Technology	Application Context	Reference(s)
Weight estimation	2D/3D cameras, computer vision, depth sensors	Non-invasive daily monitoring of growth and weight gain	[25,48,65,68,71,77]
Sound-based stress detection	Microphones, AI-based acoustic classification systems	Acoustic monitoring for signs of stress, pain or disease	[28,45,61]
Behavior monitoring	Accelerometers, microphones, posture/sound detection	Detection of stress, illness and abnormal behavior patterns	[27,28,57,61,63]
Thermal status/fever detection	Infrared thermography, thermal cameras, ear-based sensors	Continuous monitoring of body temperature and detection of fever or heat stress	[26,80,81,82,88,96]
Welfare optimization using early warning systems	AI algorithms with integrated video, audio and motion data, anomaly detection systems	Real-time alerts based on abnormal patterns to prevent welfare deterioration	[28,45,53,57,61,90]
Individual identification	RFID, facial recognition, biometric systems	Tracking and managing pigs at the individual level	[30,36,45,51,54]
Digital twin systems	Simulations of animal–environment interaction	Predictive models for management and scenario planning	[40,41,42,43]
Monitoring health and welfare	AI stress detection, facial expression and movement analysis	Real-time indicators of animal comfort, illness or distress	[60,62,63,73]
Locomotion analysis	Computer vision, accelerometers, pressure-sensing mats	Detection of lameness and mobility-related welfare issues	[69,74,75]
Estrus detection	Infrared cameras, temperature and ultrasonic sensors	Early and automated detection of reproductive status in sows	[83,84,94]
Feeding	Precision feeding systems, RFID-based feeders, smart feeders	Individualized feeding strategies to optimize nutrient use and minimize waste	[20,97,101,103,106]
Environmental monitoring	IoT sensors, smart ventilation, climate sensors	Adjusting housing conditions to maintain optimal animal welfare	[103,110,111]

Table 2. Precision livestock farming (PLF) technologies applied in pig production and their application in production cycles.

Parameter	Applied Technology	Application Context	Reference(s)
Weight estimation	2D/3D cameras, computer vision, depth sensors	Non-invasive daily monitoring of growth and weight gain	[25,48,65,68,71,77]
Sound-based stress detection	Microphones, AI-based acoustic classification systems	Acoustic monitoring for signs of stress, pain or disease	[28,45,61]
Behavior monitoring	Accelerometers, microphones, posture/sound detection	Detection of stress, illness and abnormal behavior patterns	[27,28,57,61,63]
Thermal status/fever detection	Infrared thermography, thermal cameras, ear-based sensors	Continuous monitoring of body temperature and detection of fever or heat stress	[26,80,81,82,88,96]
Welfare optimization using early warning systems	AI algorithms with integrated video, audio and motion data, anomaly detection systems	Real-time alerts based on abnormal patterns to prevent welfare deterioration	[28,45,53,57,61,90]
Individual identification	RFID, facial recognition, biometric systems	Tracking and managing pigs at the individual level	[30,36,45,51,54]
Digital twin systems	Simulations of animal–environment interaction	Predictive models for management and scenario planning	[40,41,42,43]
Monitoring health and welfare	AI stress detection, facial expression and movement analysis	Real-time indicators of animal comfort, illness or distress	[60,62,63,73]
Locomotion analysis	Computer vision, accelerometers, pressure-sensing mats	Detection of lameness and mobility-related welfare issues	[69,74,75]
Estrus detection	Infrared cameras, temperature and ultrasonic sensors	Early and automated detection of reproductive status in sows	[83,84,94]
Feeding	Precision feeding systems, RFID-based feeders, smart feeders	Individualized feeding strategies to optimize nutrient use and minimize waste	[20,97,101,103,106]
Environmental monitoring	IoT sensors, smart ventilation, climate sensors	Adjusting housing conditions to maintain optimal animal welfare	[103,110,111]

5. Conclusions

Worldwide, the further development of the PLF system is leading to improved housing conditions for pigs and higher product quality, lower carbon production per animal and significant energy savings in terms of the intake, feed conversion and growth. This review provides insights into the integration of different PLF technologies in pig production cycles and addresses the impact of novel technologies on environmental sustainability in pig production. This progress should be accompanied by the continuous training of farmers to make the applications of new technologies clearer and more accessible to them. The increasing development of digital technologies is changing pig production, making it sustainable and having a positive impact on pig welfare, health and productivity. Real-time data collection and analysis during the production process lead to early decisions that lead to better management and improved production results. With the joint collaboration of farmers, researchers and IoT experts, the application of these novel technologies will be practical and cost-effective, ensuring efficiency and long-term sustainability.

Climate change and global warming present significant challenges for pig production. As the climate, environment and pig production are interdependent, the application of PLF systems helps to improve the overall environmental performance of farms. The focus of PLF systems in pig production is on production efficiency and ensuring the ultimate profitability through labor savings, which encourages the further adoption and application of these systems in intensive pig production. Future research should focus on the introduction of PLF systems to small-scale farms and extensive systems where the implementation of this technology is still limited. By reducing equipment costs and developing models suitable for small producers, this technology should be made accessible to a larger number of producers. The integration of renewable energy sources and the principles of a circular economy with PLF systems will lead to an improvement in the overall sustainability of pig production.