In recent years, researchers have been trying to build fully automated wildlife detection techniques using modern tools and techniques to prevent human-animal conflict. Habitat destruction due to construction of roads and railways in forest areas has increased human-animal conflict, resulting in loss of both human and animal life. To avoid this conflict and maintain an ecological balance between humans and animals, wildlife detection and early warning systems are used. Manual surveillance from watch towers and traditional systems like electric fencing and trenches to avoid human-animal conflict is often unreliable and requires human interference at all times. With the emergence of new technologies such as Artificial Neural Network, Internet of Things and the availability of high speed internet connection via 4G and low-cost sensors and computers, building reliable automatic wildlife detection and warning systems which require minimum human interference is now possible. In this paper, a comparative study on the existing research papers dealing with fully automatic or semi-automatic wildlife detection techniques using modern technologies has been done. This paper attempts to explain in detail the key points and underlying technology used in the existing research papers in the concerned domain and also find the research gap if any.