In recent years, animal well-being in industrial slaughterhouses has become a significant concern for consumers, farmers, and meat producers. Different groups have different interpretations of animal well-being. For the majority of consumers, animal well-being is highly influenced by their values and experiences. Meat producers are interested in the stress animals endure because it affects meat quality.Pigs that arrive at slaughterhouses are more sensitive than usual for several reasons. In some cases, pigs are transported for long distances. Not all animals are used to transportation. Upon their arrival, it is common to mix pigs from different farmers in one area. Such mixing can cause fights between pigs, which can lead to additional stress or the animals being harmed. The unfamiliar environment also increases the animals’ stress levels. In some industrial slaughterhouses, up to 62,000 pigs per week are handled. Ensuring the well-being of such large numbers of pigs using only personnel is a complicated task. Video surveillance of humans has been widely used to ensure safety and order in multiple situations. Methods have been developed to detect individual actions or abnormal behavior in small groups and dense crowds. In recent years, surveillance has also been used to monitor animals. Research has mainly focused on monitoring laboratory animals and farm animals. In both cases, animals are usually in constrained environment and cameras are used to cover all areas where animals are present. To obtain better results, non-intrusive markers or extracted features are used for tracking. Laboratory environments can be highly controlled; thus, no light and shadow noise are present in videos.In slaughterhouses, the main focus is on monitoring large groups of animals in locations where additional markers cannot be used and pigs can leave or enter the surveilled area. In addition, pigs have a specific walking pattern; thus, motion analysis is not straightforward. The first aim of this thesis is to monitor the movement of pigs without using any additional markers or feature extraction in an unconstrained environment.In video surveillance, the behavior of humans and animals is monitored based on extremes: event is present/event is not present, objects behave normally/objects behave abnormally, action 1/action 2/action 3, etc. In nature, the motion of humans and animals is continuous with transitions from one action to another. The second aim of this thesis is to propose a method to monitor motion as a continuous process using common classification methods.