While railroad trespassing is a dangerous activity with significant security and safety risks, regular patrolling of potential trespassing sites is infeasible due to exceedingly high resource demands and personnel costs. There is thus a need to design an automated trespass detection and early warning prediction tool leveraging state-of-the-art machine learning techniques. Leveraging video surveillance through security cameras, this thesis designs a novel approach called ARTS (Automated Railway Trespassing detection System) that tackles the problem of detecting trespassing activity. In particular, we adopt a CNN-based deep learning architecture (Faster-RCNN) as the core component of our solution. However, these deep learning-based methods, while effective, are known to be computationally expensive and time consuming, especially when applied to a large amount of surveillance data. Given the sparsity of railroad trespassing activity, we design a dual-stage deep learning architecture composed of an inexpensive prefiltering stage for activity detection followed by a high fidelity trespass detection stage for robust classification. The former is responsible for filtering out frames that show little to no activity, this way reducing the amount of data to be processed by the later more compute-intensive stage which adopts state-of-the-art Faster-RCNN to ensure effective classification of trespassing activity. The resulting dual-stage architecture ARTS represents a flexible solution capable of trading-off performance and computational time. We demonstrate the efficacy of our approach on a public domain surveillance dataset.

Inom byggbranschen finns det problem med att personer kommer in på arbetsområdet där de kan komma till skada. Detta problem kallas problemet med tredje man på arbetsplats och byggföretagen jobbar idag på att hitta en lösning till detta. I examensarbetet ligger fokusen på att identifiera och kartlägga hur incidenterna går till och om det är möjligt att förebygga dessa. I rapporteringsdatabasen BIA har 1000 incidenter undersökts, kategoriserats och presenterats. Detta för att hitta samband och beteendemönster. Intervjuer har gjorts med branschaktiva och enkäter har skickats ut till insatta i ämnet för att inhämta kunskap och förståelse av problemet. De studerade incidenterna som undersökts kommer från Skanskas arbetsplatser. Incidenter som inträffat på andra ställen än Skanska Sverige undersöks inte. Studien undersöker inte om det är möjligt att förhindra skador på tredje man. Studien har visat att det finns vissa beteendemönster i hur och varför tredje man beträder arbetsplatser. Dessa beteendemönster kan delvis förklaras av hur och när folk tar sig till skolor och jobb. Studien visar också att i vissa fall har tredje man kommit in av misstag då avstängningar är otillräckliga. Material från intervjuer och enkäter stödjer också detta resultat. I dessa intervjuer och enkäter framgår också att det finns ett stort problem av bristande respekt hos tredje man vilket leder till aktiva beslut att beträda arbetsplatser. Slutsatsen som kan dras av examensarbetet är att problemet med tredje man på arbetsplatser är omfattande och det finns många orsaker till problemet. Det framgår också att problemet med tredje man existerar i betydande grad bland fotgängare, cyklister och bilister. Det som kan konstateras är att om problemet helt ska förhindras krävs åtgärder på många olika sätt, både för att stoppa de aktiva intrången och förhindra att personer kommer in av misstag eller missförstånd. / In the construction industry today, there is problems related to the public entering worksites where accidents might happen that also lead to injuries. This is called the problem with third person at worksites and today, companies do their best to find a solution for this. In this examination thesis the focus is to identify and map how these incidents occur, and if it is possible to prevent them. In the report database BIA, 1000 incidents has been studied, categorized and presented. This has been done to find similarities and behavioral patterns. To collect information and understanding about this issue, interviews and surveys has been done with people working in the construction industry. The incident reports that has been studied comes from Skanska’s worksites in Sweden. Incidents that has occurred at other work sites has not been studied. The study does not analyze if it is possible to avoid injuries to unauthorized people entering the work sites. The study shows there is signs of behavioral patterns concerning the public and how they end up within worksites. This can partly be explained in which time and way people travel between home and schools or work. The study also shows that in some cases third person has entered the worksite by mistake when fences and gates has been insufficient. The interview material and surveys also support this result. In these interviews and surveys it’s also apparent that there is a big issue concerning peoples lack of judgement, leading to decisions to enter construction sites. What can be summarized by this examination thesis is that the problem with unauthorized entry by third person at worksites is a major issue and there seems to be many causes to the problem. It also appears that the issue widely affects pedestrians, cyclists and motorists. What can be said about this is that if the problem is to be completely prevented, large effort needs to be applied in many ways, both to prevent active intrusion and to prevent people trespassing by accident.

To gather together is a natural human activity shared by all people. The majority of these activities take place without the involvement of the government, and is of no interest to the law. In South Africa, the right to assemble peacefully, to demonstrate, to picket or to present petitions, is protected in the Constitution of the Republic of South Africa, 1996. When people gather, be it peaceful or violent, participants run the risk of being arrested for committing offences. The way the government of the day reacts to gatherings influence the policing, prosecution and adjudication of offences arising from the right to gather. Current legislation and common-law offences utilised to curb disorder in South Africa are measured against international and regional case law and guidelines. Most of these case law and guidelines linked to international and regional instruments are similar in many respects, and can be deemed as universally acceptable. It is proposed that the government revisits the mixture of current offences utilised by the prosecution during dissent, public violence or protest action, and that specific public order offences are created, providing for specific unlawful conduct with corroborating sentences. Police powers must furthermore be clearly defined to strengthen the hand of the police to secure law and order, serve as guarantee for the rights and freedoms of everyone, and to create legal certainty. The government must organise applicable public order offences in a single public order act. Legislation applicable to public order must be accessible and easily understandable since protest may be the only avenue for a member of the public to bring his or her plight under the attention of the government. Existing guidelines from applicable international and regional instruments which guide and monitor executive conduct must be included since these guidelines qualify as public order offences. / Criminal and Procedural Law / LL. D. (Criminal and Procedural Law)