This work presents the results of a survey through the analysis of studies published between January 2017 and May 2021, aiming to compose a broader view of the state of the art in the field of animal detection and classification using computer vision technologies in urban environments, and also the majors researches gaps available to address. We conducted an automatic search through two digital knowledge bases identifying 146 studies in the subject, among them 20 were selected for our analysis and data extraction. Further, the 20 studies were classified into 6 categories: (i) studies using SVM, (ii) studies using HOG, (iii) studies using SIFT, (iv) studies using PCA, (v) studies using CNN, and (vi) DFDL. As a result, it can be noted that the use of CNN is predominant concerning other approaches and that there are also combinations to improve the accuracy of classification models. In conclusion, it is possible to observe that the state-of-the-art approaches have been used in different situations, however, in the context of animal detection and classification in intelligent urban environments, there is still a lack of specific architectures to improve results.

The increase in computational technologies for means of transport, especially vehicles, has provided great benefits through Intelligent Transport Systems (ITS). Drivers, passengers, and pedestrians rely on computer applications that aim to protect human life, including agility in handling emergencies, improvements in traffic, and even leisure and entertainment resources. Communication and data exchange are from vehicles to vehicles (V2V) and from vehicles to road infrastructures (V2I) being carried out through the architecture of the vehicular ad hoc network (VANET). However, this type of network differs from traditional ones, as it operates in a highly dynamic environment, originated by the rapid mobility between its nodes and with short connection intervals. Wireless vehicle communication adopts the IEEE 802.11p standard, allowing vehicles to operate outside a basic set of services. Given these characteristics, numerous threats, vulnerabilities, and denial of service attacks can occur. Prioritizing the safety of life and protecting VANET against this type of attack, a security mechanism is proposed. The mechanism works to detect anomalies through a simple and robust statistical model in the search for extreme values (outliers). Median Absolute Deviation detects large amounts of MAC frames and ARP requests, characteristics of DoS / DDoS from malicious vehicles. Through extensive stages of simulations using the NS-3 and SUMO simulators, the mechanism showed excellent efficiency in detection rates and minimum rates of false positives and false negatives.