Convolutional Neural Networks for Individual‑identification of Wild Asian Elephants

Visual monitoring of wild animals has been modernized over the years through technologies such as high definition photography and camera trapping[1]. Researchers can now document populations, movement, and behavior of certain species using large volumes of data over longer periods[1,2]. These tools aid population research, but photo-identification still relies on our ability as humans to distinguish certain features of individuals[3]. Because identification remains a manual task, extracting information from visual data can be expensive and time-consuming[4]. (Read our previous blog on UWERP’s EARS – Elephant Attribute Recording System – IDs Database here).

MSc student, Elgiriyage de Silva, at the University of Colombo, is lead author on their recently published CNN study, alongside others including Dr. Shermin de Silva and Udawalawe Elephant Research Project’s (UWERP) Research Supervisor, T.V. Kumara. The study made use of CNN in order to determine the feasibility of such technology to identify Asian elephants, and used 10 years of labeled photographs of wild Asian elephants collected by the UWERP study. The researchers considered full body, face, and ears as three points for individual identification. Two techniques namely Training from Scratch (TS)  and Transfer Learning (TL), which made use of a pre-trained model, were applied to five CNN models: Xception, Inception V3, VGG16, ResNet50, AlexNet. These models were evaluated for their efficiency in correctly identifying an individual as the top candidate or including the correct individual among the top five possible candidates[3]

Caption: Photos of male 006 showing all three types of images used in the study – body, face and ears. Clearly photos of even the same individual can look very different based on angle, lighting and movement!

The results of the study revealed that the VGG16 model trained by TL yielded the highest prediction accuracy at 21.34% and 42.35% for the top candidate when using the full body and faces dataset, respectively. The best-performer was the Xception model, trained on ears with the TS technique, which returned a prediction accuracy of 89.02% for identifying the correct individual, and 99.27% for including it among the top five candidates. Interestingly, ears turned out to be the most identifiable features for the CNN algorithm, just as they are for humans. With these impressive results, de Silva and colleagues (2022) concluded that it is possible to accurately automate the identification of Asian elephants, but with certain caveats[3]

To solve this predicament, advancements using Artificial Intelligence, more specifically machine learning (ML) techniques, have facilitated computer-mediated identification among animals. One of the well-known systems used for image classification under ML is Convolutional Neural Networks (CNN). Likened to the function of neurons in the human brain, CNN are potent artificial intelligence technologies that utilize deep learning to execute generative and performative functions such as image classification and object detection. While this technology has been proven useful among animals with unique skin or coat patterns such as whale sharks and tigers, identification of species that lack distinguishing features, such as Asian elephants, remains a challenge.

Photo caption: Amali was the first individual catalogued by the Udawalawe Elephant Research Project, estimated to be 50 years old by 2020. The front-view showing both ears exhibits the features long used by humans, and now by computers, for discriminating among individuals. Asian elephant ears fold along the upper (primary) crease, which can be forward, upright, or sloped back, as well as the side (secondary), which can flap forward or backward. Amali’s two ears are asymmetric, which is also common. Together with 3-dimensional ear shape, damage, and vein patterns, the number of possible feature combinations is large enough to be individually unique within a population. As with all female Asian elephants, she lacks tusks and therefore these cannot serve for individual identification. Photo taken in Udawalawe National Park, by Shermin de Silva.

Since automated identification requires training data, the study notes that the first drawback in the system’s efficiency would be the availability of photos for model training per individual. As the target populations are found in the wild, it would take time to build an extensive library of labeled photographs which will still include initial individual identification by a human, thus defeating the purpose of automation[3]. (See our new C.O.O.’s blog here and here from her PhD fieldwork back in 2011 and 2013 where she continued to unravel identity puzzles for the calves she was studying in Udawalawe NP!) Furthermore, field conditions may restrict collection of full body and face photos through obstructions such as vegetation and foliage. Aside from this, Asian elephants’ features change over time, including changes brought about by aging or injuries, so the system would require periodic updating and re-training.

These limitations raise the question: is it practical to use CNN techniques for Asian elephants in the wild?

According to de Silva et al. (2022), this technique is more feasible for long-term monitoring due to the amount of time and resources needed to train, test, and maintain the system. As long as a population is well-cataloged with sufficient images, automated identification can save time by narrowing down potential candidates. Since the ideal situation is where there are are a large number of high quality photographs for each individual, this technique can aid wildlife stakeholders in activities such as tracking rehabilitated and translocated individuals and countering illegal trade involving falsified  documentation[5,6].

References:

1Kays, R., Tilak, S., Kranstauber, B., Jansen, P. A., Carbone, C., Rowcliffe, M. J., Fountain, T., Eggert, J., & He, Z. (2010). Monitoring wild animal communities with arrays of motion sensitive camera traps. ArXiv:1009.5718 [Cs]. http://arxiv.org/abs/1009.5718

2O’Brien, T. (2011). Camera Traps in Animal Ecology (pp. 71–96). https://doi.org/10.1007/978-4-431-99495-4_6

3de Silva, E. M. K., Kumarasinghe, P., Indrajith, K. K. D. A. K., Pushpakumara, T. V., Vimukthi, R. D. Y., de Zoysa, K., Gunawardana, K., & de Silva, S. (2022). Feasibility of using convolutional neural networks for individual-identification of wild Asian elephants. Mammalian Biology. https://doi.org/10.1007/s42991-021-00206-2

4Norouzzadeh, M. S., Nguyen, A., Kosmala, M., Swanson, A., Palmer, M. S., Packer, C., & Clune, J. (2018). Automatically identifying, counting, and describing wild animals in camera-trap images with deep learning. Proceedings of the National Academy of Sciences, 115(25), E5716–E5725. https://doi.org/10.1073/pnas.1719367115

5Menon V, Tiwari SK (2019) Population status of Asian elephants Elephas maximus and key threats. Int Zoo Year 53(1):17–30. https://doi.org/10.1111/izy.12247

6Prakash TSL, Indrajith WU, Aththanayaka A, Karunarathna S, Botejue M, Nijman V, Henkanaththegedara S (2020) Illegal capture and internal trade of wild Asian elephants (Elephas maximus) in Sri Lanka. Nat Cons 42:51

The Global Goals and Asian Elephant Conservation

The 17 Sustainable Development Goals provide a blueprint to achieve a better and more sustainable future for people and the planet. In celebration of Earth Month, we broke down all 17 Global Goals to discover how each relates back to our mission to protect and conserve Asian elephants and their habitat.

Global Goal 1: No Poverty

In developing countries where elephants roam wild, like Sri Lanka, poverty and elephants can become intertwined. Small farmers can lose their entire livelihood overnight from an elephant raid, and an 8,000 pound animal walking through a farm can destroy everything in its path.

Finding ways for farmers to make a living alongside Asian elephants is key to the survival and success of both elephants and people. Our Coexistence Project studies both sides to develop innovative ways that farmers can maintain a steady income while living peacefully alongside wild Asian elephants.

Continue reading

The Sumatran Elephant: Human-Elephant Conflict, Habitat Use and Home Ranges

By Gaius Wilson

The Sumatran elephant (Elephas maximus sumatranus), found only on the island of Sumatra in Indonesia, is critically endangered according to the International Union for the Conservation of Nature (IUCN). The population is decreasing with approximately 1500 elephants left in the wild in fragmented populations. Deforestation, loss of habitat and poaching for ivory are amongst the major threats to the survival of this species.

The Leuser Ecosystem (which forms a significant part of the UNESCO World Heritage site ‘Tropical Rainforest Heritage of Sumatra’) and Ulu Masen in Aceh, Sumatra are a stronghold for the critically endangered Sumatran elephant and other critically endangered wildlife (e.g. orangutans, rhinoceros, and tigers). Both Leuser and Ulu Masen are essential for the survival and conservation of the Sumatran elephant, but much of their habitat falls outside the protected areas and in the most threatened lowland forests, creating elephant human contact. This makes it critical that effective mitigation strategies are developed that take into account elephant behaviour and the use of technology such as early warning systems to reduce conflict with the local communities.

Continue reading

Can Asian elephants use water as a tool?

By Dr. Lisa P. Barrett

Asian elephants at the Oklahoma City Zoo

The floating object task is a puzzle that comparative cognition researchers present to animals (including humans) to study the evolution of cognitive abilities, like cause and effect understanding and the ability to use water as a tool. To solve the task and retrieve the floating reward inside, you must add water to a tube to raise the water level and reach the reward. Some primates, like orangutans have been able to solve the task by carrying water in their mouths from a drinker and spitting it into the tube to reach a peanut.

My colleague, Dr. Sarah Benson-Amram, and I presented this puzzle to elephants for the first time (Barrett & Benson-Amram, 2020). We wanted to see if elephants’ unique trunk morphology would make them well-equipped for the floating object task. Since they spray water for bathing, and hold water in their trunk as a vessel for bringing it to their mouth, we predicted that they would be up for the task. We collaborated with the National Zoo and the Oklahoma City Zoo to carry out this research. We used a tube filled about 1/3 of the way with water, baited with a floating marshmallow. As is often the case with animal research, things did not go as we expected.

Continue reading

Deja Vu at Dahaiyagala

by SdS

Elephants at the Pokunutenna reservoir.

When we saw the destruction, we felt that strange sensation of history repeating. We’d been here before, a little over ten years ago.

Dahaiyagala sanctuary is a little-known, nondescript little strip of forest north of Udawalawe National Park, one of the two official “corridors” that supposedly links the park to the outside world. It is supposed to lead to another forest area, which conservationists and wildlife authorities refer to as Bogahapattiya. It also borders Pokunutenna village, a hotbed of unrest with respect to human-elephant conflict. Dahaiyagala represents unfinished business to the various parties, in very different ways.

Continue reading

A Christmas Wish

By USW

Back in 2004, a male calf was born to the young female elephant, 458. The calf had baby tusks and this made his birth all the more special. He was named Therapuththa, or T458. We named him Therapuththa but labeled him as T458, following his mother’s ID, this is because it helps us keep track of him in the future when we catalog his movements, features and characteristics. His name, Therapuththa, was taken from ancient history: it is said that King Dutugemunu had 10 giant warriors protecting him and Therapuththa was one of the most loyal and endearing ones. The mother, 458, was often spotted with two other female elephants, who we have labelled as 040 and 041. After tracking this herd for so many years, we have unfortunately not seen any calves after Therapuththa. So he was quite special, besides being a tusker, which is rare in Sri Lanka.

Continue reading

Bees are Helping Thailand’s Elephants and Farmers to Peacefully Coexist – Bee Fences in Asia Part 2

A new study brings hope for reducing conflicts between elephants and farmers in Thailand

Guest post by Antoinette van de Water

Beehive fences in Thailand. Photo: BTEH

Kaeng Hang Maeo district in Eastern Thailand is in an area of high human-elephant conflict. A herd of about 70–80 elephants lives between the protected areas and agricultural land, causing damage to crops almost on a nightly basis. Over four years ago, Bring The Elephant Home (BTEH) and the Phuluang Wildlife Research Station started a joint project to evaluate the effectiveness of beehive fences in deterring Asian elephants, under supervision of Dr. Lucy King. We set up a pilot beehive fence around a subsistence farm surrounded by elephant habitat and installed camera-traps to record the elephants’ reactions to the bees, which belong the species Apis mellifera, or European honeybee.

Continue reading

Can bees help farmers in Sri Lanka deter elephants from their crops? Bee fences in Asia – Part 1.

The tiny bee vs. the world’s largest land mammal…

Guest post by Kylie Butler

Elephants outside Wasgamuwa National Park / Apis cerana bees being transferred into a hive (Photos: Kylie Butler)

Over a decade ago now, Dr. Lucy King developed the beehive fence as an elephant deterrent, capitalising on a then-recent discovery that African elephants avoided African honeybees (Vollrath & Douglas-Hamilton, 2002). The beehive fence is a relatively simple, inexpensive deterrent, aiming to be a tool that communities can use independently following set-up. The basic premise is that a series of beehives surround an area to be protected from elephants, and if elephants attempt to enter, they will disturb the beehives, causing the colonies to swarm (refer to King et al. 2009; 2017 for more details). It should come as no surprise, that the success of multiple beehive fence trials in Africa, led to a curiosity as to whether this technique could also help Asian communities experiencing comparable levels of crop-raiding.

Continue reading

Responsible Tourism & Ethical Elephant Experiences

Mother and calf - ethical tourismOn World Elephant Day, Trunks & Leaves is challenging travelers and travel companies alike to commit to responsible tourism practices when it comes to viewing and interacting with Asian elephants.

 For the first time in recent history, the world has slowed down, the travel industry is on hold, and humankind has a chance to reflect on the way we’re doing things and how we can improve in the future – for both humans and wildlife.

Continue reading

Explosive Food and What it Tells Us About Ourselves

By SdS

Injured elephant in water.

Photo by Mohan Krishnan of injured elephant in the water.

Somewhere, there is a hungry elephant, following her nose, wandering an ever-diminishing forest in search of food. She ventures to her usual places, finds them lacking. She wanders further from where she feels safe, considering what she may find closer to the villages nearby.

Somewhere there is a hungry person. Perhaps a farmer, perhaps a hunter. He is looking to drive away pests from his land, or maybe to earn a bit of money from bushmeat. He selects a large fruit or vegetable, say a pumpkin or pineapple. He hollows it out, hides an improvised explosive inside, leaves it where some animal will find it.

We know what happens next. Continue reading