Animal Detection and Counting using Single Shot Multibox Detector

Abstract

In this paper we combine image processing and machine learning to observe animals inside the homestead while not upsetting them, is plausible exploitation camera lodgings system, that might be a method to check life exploitation by precisely activated cameras and produces decent volumes of data. Nonetheless, camera lodgings gather pictures for the most part end in low picture quality and incorporate a lot of bogus positives (pictures while not creatures), that ought to be located before the post-procedure step. This paper presents a two channelled perceiving residual pyramid networks (TPRPN) for camera tempt pictures complaint. Our TPRPN model takes care of creating high-goals and top-notch results. So as to deliver enough local data, this research will in general concentrate profundity signal from the underlying pictures and utilize the two-diverted seeing model as a contribution to instructing our systems. At last, the anticipated three layer lingering squares figure out how to consolidate all the information and produce full size identification results. Furthermore, this research builds another top-notch dataset with the assistance of Wildlife Thailand's Community and mammal creature Organization.