Fossil Finder

Fossil Finder - the next stage. April 2018 Update

We have been astounded by the success of Fossil Finder over the course of the project so far. We cannot thank all of the wonderful citizen scientists enough for their contributions – there is no way that we could have done this project without your help. In summary, to date we have had 59,000 subject images that have been studied by a combination of more than 5,600 registered citizen scientists and over 5000 anonymous sessions. In total, greater than 675,000 classifications of potentially interesting objects have been achieved since the Fossil Finder webpage was launched in 2015. However, this is not the final aim of the project and thus in August 2016 project members were able to successfully ground truth several of the finds that had been located online by our citizen scientists, by physically visiting the locations of a limited number of the classified objects. This confirmed the capability and usefulness of the Fossil Finder framework we have produced, but it is not feasible to visit the location of every single object classified by the citizen scientists.

It has been just over two years since that launch and we are now getting to grips with the dataset that has been produced. Within this data we are finding some patterns that are interesting and useful, and demonstrate the potential for future research of this kind.

We are now moving along to a second phase of the citizen science data analysis, where we need you to evaluate the quality and precision of the initial data classification. The first batch of data has been studied using clustering algorithms to identify hotspots of consensus among the citizen scientists’ classification data. To evaluate the quality of this higher level data we are asking volunteers to evaluate if the clusters have correctly identified and labelled objects of interest. There is now a new live workflow online that will allow people to take part in this new effort.

We're really excited to see what you have helped us to find!

At the moment you will be able to see images which show classification clusters highlighted with a blue ring. The clusters contain two types of identification points, where citizen scientists have either classified the object as a bone/tooth (blue circle) or as 'might be something'(red triangle). We need your help to determine whether there is a consensus that these can indeed be identified as bone/tooth or not. Please feel free to take a look at these images and help us refine our analytical procedures in order that we might clean up our datasets and be more certain in the classifications made.

Check out our blog for updates as we progress.!