Machine learning algorithm for accelerometer/gyroscope animal movement data.
Budget: $250 – $750 USD
We are looking for a person experienced in ML building, preferably in Python or R, with experience in movement data analysis to work with already collected data.
We also look ESPECIALLY for those who love animals and would like to participate in such an activity and save the wildlife, YES, even in such a way as building AI!
The tasks are:
- To build a supervised ML algorithm to classify different movement and behaviour types from already collected data.
- To be able to assign particular types of behaviour in training datasets.
- To advise the best practices and explain the mathematical methods used in the algorithm.
- To write the explanation and methodology behind the models used in the algorithm.
P.S. For those who are really interested and are looking for more details
We have non-stop 20 Hz accelerometer and gyroscope movement data from 10 straight days from 10 animal-borne collars.
The collars were fitted on 10 Barbary sheep females in captivity (photo is attached).
We also have videos of all their activity during those 10 days.
And some activity already is marked up for the test sets of ML.
The aim is to train the model to "see" later such types of activity like walking, running, resting, feeding, rumination etc.
This ML learning will be later used as a part of a wildlife monitoring tools for a better understanding of protected species behaviour, subsequently - better ways of saving and protecting them.
We also look ESPECIALLY for those who love animals and would like to participate in such an activity and save the wildlife, YES, even in such a way as building AI!
The tasks are:
- To build a supervised ML algorithm to classify different movement and behaviour types from already collected data.
- To be able to assign particular types of behaviour in training datasets.
- To advise the best practices and explain the mathematical methods used in the algorithm.
- To write the explanation and methodology behind the models used in the algorithm.
P.S. For those who are really interested and are looking for more details
We have non-stop 20 Hz accelerometer and gyroscope movement data from 10 straight days from 10 animal-borne collars.
The collars were fitted on 10 Barbary sheep females in captivity (photo is attached).
We also have videos of all their activity during those 10 days.
And some activity already is marked up for the test sets of ML.
The aim is to train the model to "see" later such types of activity like walking, running, resting, feeding, rumination etc.
This ML learning will be later used as a part of a wildlife monitoring tools for a better understanding of protected species behaviour, subsequently - better ways of saving and protecting them.