Deep CNN Based Automatic Detection and Identification of Bengal Tigers

A system for individual identification of The Royal Bengal Tigers (Panthera tigris) is absolutely necessary not only for monitoring the population of tigers but also for saving the precious lives of those workers whose job is to count the exact number of tigers present in a particular region like Sundarban in West Bengal, India. In this paper, a solution has been proposed for individual identification of Bengal Tigers using an autonomous/manually controlled drone. In the proposed system, the drone camera will search for the tigers using a Tiger Detection Model and then the flank (the body part which contains the stripes) of the detected tiger will be passed through a Fine-tuned state-of-art network. The system based on deep CNN will detect the uncommon features for individual counting of the tiger in a particular forest. The proposed system will enhance the accuracy of tiger detection technique that will be followed by the human experts. It also reduces the risk of accidents relating to animal attacks.

[1]  Santosh Kumar,et al.  Cattle Recognition: A New Frontier in Visual Animal Biometrics Research , 2020, Proceedings of the National Academy of Sciences, India Section A: Physical Sciences.

[2]  Janko Calic,et al.  Tracking Animals in Wildlife Videos Using Face Detection , 2004, EWIMT.

[3]  Taghi M. Khoshgoftaar,et al.  A survey on Image Data Augmentation for Deep Learning , 2019, Journal of Big Data.

[4]  Y. Jhala,et al.  Identification of individual tigers (Panthera tigris) from their pugmarks , 2005 .

[5]  Yulong Zhang,et al.  Pet Face Detection , 2018, 2018 Nicograph International (NicoInt).

[6]  Andrew Zisserman,et al.  Very Deep Convolutional Networks for Large-Scale Image Recognition , 2014, ICLR.

[7]  Peijiang Chen.  Moving Object Detection Based on Background Extraction , 2009, 2009 International Symposium on Computer Network and Multimedia Technology.

[8]  Camille Couprie,et al.  Learning Hierarchical Features for Scene Labeling , 2013, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[9]  G. Seber,et al.  Estimating Animal Abundance: Review III , 1999 .

[10]  Xudong Jiang,et al.  Fingerprint minutiae matching based on the local and global structures , 2000, Proceedings 15th International Conference on Pattern Recognition. ICPR-2000.

[11]  Mauro Castelli,et al.  How Deeply to Fine-Tune a Convolutional Neural Network: A Case Study Using a Histopathology Dataset , 2020, Applied Sciences.

[12]  Michał Grochowski,et al.  Data augmentation for improving deep learning in image classification problem , 2018, 2018 International Interdisciplinary PhD Workshop (IIPhDW).

[13]  T. Burghardt,et al.  Real-time Face Detection and Tracking of Animals , 2006, 2006 8th Seminar on Neural Network Applications in Electrical Engineering.

[14]  S. Thorpe,et al.  A Limit to the Speed of Processing in Ultra-Rapid Visual Categorization of Novel Natural Scenes , 2001, Journal of Cognitive Neuroscience.

[15]  James D. Nichols,et al.  Estimation of tiger densities in the tropical dry forests of Panna, Central India, using photographic capture–recapture sampling , 2004 .

[16]  Geoffrey E. Hinton,et al.  Deep Learning , 2015, Nature.

[17]  Graeme Caughley,et al.  Sampling in Aerial Survey , 1977 .