Gaja-Mithuru: Smart Elephant Monitoring and Tracking System

Toward Long Range Detection of Elephants Using Seismic Signals: A Geophone-Sensor Interface for Embedded Systems

“…Among these studies, a successful approach was taken by Parihar et al in [50], utilizing a highly sensitive geophone (85.8 V/m/s) with a single-stage low-pass RC filter and 16-bit ADC, achieving a tested range of up to 40 m. However, there was a significant decrease in reported accuracy concerning distance (at 20-40 m range) due to high-frequency background noise. Similarly, [51], [55], and [52] present IoT-enabled systems for HEC using seismic signals, and [52] shows a successful study using seismic signals to detect elephants while employ- ing an amplifier and a bandpass filter. Notably, the authors of all studies highlight the need for further improvement in analog signal processing for better accuracy.…”

“…Compared to two methods: sound card and seismometer, for detecting elephant locomotion using seismic waves, the use of generic embedded systems is not only cost-effective and easier to implement in rough terrains but also customizable according to the requirement [50], [52]- [55]. Among these studies, a successful approach was taken by Parihar et al in [50], utilizing a highly sensitive geophone (85.8 V/m/s) with a single-stage low-pass RC filter and 16-bit ADC, achieving a tested range of up to 40 m. However, there was a significant decrease in reported accuracy concerning distance (at 20-40 m range) due to high-frequency background noise.…”

“…Notably, the authors of all studies highlight the need for further improvement in analog signal processing for better accuracy. Moreover, as indicated in Table 1, despite all the studies [50], [52]- [55] concluding that the use of embedded systems together with a geophone demonstrated promising accuracies for non-invasive elephant detection, none of the studies reported a detection range beyond 40 m. Unfortunately, apart from [54], all the studies failed to provide details regarding the gains employed in their systems. Similarly, none of the studies discusses the impact of gain adjustments on the detection range, a crucial factor that hinders the implementation of long-range elephant detection.…”

“…The highlighted limitations, i.e. the requirement for improved amplification under different gain settings and noise filtration to enhance the elephant detection range (especially over long distances) [50]- [52], the high cost of equipment and the use of sophisticated technologies [48], [49], the lack of rigorous testing in tropical terrains with adequate parameters (for noise filtration and signal amplification gains) , and the limitation elucidated when interfacing of geophone to an embedded system [50]- [52], incur the necessity for a sophisticated geophone-sensor interface to improve the accuracy of detecting elephants over long distances. Hence, this study proposes a geophone-sensor interface with the aim of amplifying signals by employing adjustable gains and eliminating seismic noise for effective and accurate longrange detection of elephants based on their characteristics in tropical terrains.…”

Elephants and algorithms: a review of the current and future role of AI in elephant monitoring

“…Non-ML techniques have been employed to detect elephant presence from seismic data for censusing purposes, achieving 85% accuracy [154] and continuous wavelet transforms reached 90% accuracy in detecting forest elephants [155]. Within the realm of ML, elephant calls have been classified from seismic measurements using support vector machines (SVMs) with 73% accuracy [153], neural networks with 87% accuracy [156] and CNNs attaining 80–90% accuracy up to 100 m away [157].…”