Extraordinary Fantastic News
BeeScanning will be fully funded!
We have been selected by the Swedish Agricultural Department external assessment board "As a good innovation project that fits to the purpose of the European Innovation Program" We are asked to complete certain budget posts but are confident this means we can fund the project.
On top of this we today received approval from the Swedish National Honey Board and will be able to fund our marketing and information campaign with more than USD 20 000.
This all means we now have given "GO" to developers to fully concentrate on BeeScanning from the first of June.
To give you a clue that will cost the project USD 500/hour!
We also believe and have strong indications that you the Backers on Kickstarter has been an important motivator in the decision process. Authorities mentions the impact of the Crowd Funding success! Thank you backers!
New revolutionary tool launching
For the first time beekeepers around the world can collaborate combating Varro destructor thanks to Kickstarter fundings.
BeeScanning with BeeTagTool: is building the largest database of visually tagged photographs of bees. You can help in the following ways: • Upload photographs of your bees • Inspect and visually tag features: varroa mites, deformed wings, or where the queen is. • Or help tag someone elses photos! It is easy, fun and important.
Once we have several thousands of samples of varroa mite infested bees, deformed wings, queen bees, etc, we will train an neural network to learn to identify these features automatically. The result is an app - BeeScanning - that you or anyone else can use to automatically and immediately detect these features among bees, by just taking a photograph with your mobile phone. http://beetool.beescanning.com/
So to be very clear this is not the app! This is not the neural network software! This is the tool to collect the data for training the software. And this has become possible through the pledges you have made. It is really awesome and a true Kickstart! Thank you all so much. Team/Björn
Without bees flowers will be gone
A trailer version of the video wraps it up!
New goals will enable us pushing forward!
$ 7k - we build a tool that lets anyone highlight mite infested bees in photos and start collaborating with training data
$ 15k - get statistics and graphs from the app about your colonies and infestation levels
$ 30k - use the app offline, by letting the phone itself perform all calculations
$ 50k - community features, make beekeeping digital and social
$ 75k - Support more platforms, a web interface for accessing your data and the community
$ 150k - expand the tools and the training to detect more than just varroa mites
$ 300k - long term selective breeding tools - neural net training and statistics combine in a tool that optimizes selective breeding according to your criterias
$ 600k - project can be financed in full, including all above, without depending on financing outside of Kickstarter. This means whether or not we are granted the EIP (EuropeanInnovationProgram) funding or not, the project is happening.
Millions of honeybee colonies will die each year! In the US about 44%! That's terrible - almost half of the colonies - each year! The main cause is the parasite mite - named Varroa Destructor. The name is very appropriate, it destroys the bees. In Europe almost 20% are lost yearly. My neighbour and collegue had 800 colonies and lost 500! Following winter he lost another 250, so now he keeps 50 hives.
The bees really suffer. Imagine having big brown crabs crawling your body sucking your blood!
Beekeepers around the world are struggling to save their bees. But they fail to detect the infestation in time - and so their bees dies in the winter.
Its a shame!
We can change that. Use the camera in your phone. Monitor the mite.
My name is Björn Lagerman. I have about 100 beehives in the middle of Sweden.
I have been taking pictures of my bees for years. Trying to learn more. For example visit www.fribi.se/yngelramar/
Last summer I discovered Varroa mites examining close up still images. It then struck me that if I could compare the images with results from conventional chemical monitoring we would have a new tool. And indeed there was correlation!
The idea is to let an app reveal the mites in the images. Then you decide to treat or not. The app will also provide basis for finding bees that show resistance against Varroa. This is very important! We can not depend on chemicals, we must find resistant bees. Beekeepers gets a tool and researchers gets knowledge to find varroa resistant strains of bees.
This is a concept image:
In our database we will store images and data from the beekeepers. Our image analyzing technology is based on cutting edge neural networks and deep learning.
For training our neuron nets we need lots of data and therefor we are collecting pictures and survey data from 10 000 colonies taken by beekeepers worldwide.
The correlation between images, how the bees perform, what crop they bring and how they survive is used for the software development.
We then will compare received data with our own controlled extended survey -where we use alcohol washing as benchmark to measure the infestation.
How does the neural network learn?
The neural network is actively taught by a deep learning professional who has access to a good amount of training data. In short, an image of a varroa infested bee is shown to the neural network and we let the network guess. If the network guesses wrong, we let it know. It will then adjust itself very slightly to make a more educated guess next time. By doing this several thousand times the neural network will learn very abstract features and eventually become better than a human at visually detecting mites.
That's a very simplified way of looking at it. In reality, an artificla neural network is a very complex beast that is difficult to tame and comes in many different forms and shapes. One of the first milestones is for the development team to find the way of training that is optimal for this particular task.
What if you don't find enough training images for the neural network?
Less images means the precision of the system is not optimal. We can still launch the product and use the data gathered from the app as a tool for further ongoing training of the neural network.
Is the image development team actually capable of solving this task?
Yes! The team members have over 15 years of experience with developing complex systems. The team is lead by mobile and tech industry veteran Emil Romanus. Romanus has previously been involved in mobile world-wide successes, among others the complex 3D and physics simulation game "Apparatus" in 2011. Romanus and his team has been focusing on deep learning for the past few years and have made several large investments in the hardware required for the training of neural networks.
Does the development team have experience with similar solutions?
Romanus' team has worked with several other companies and solved similar issues. Most importantly the team has worked with a british company that identifies and reports the occurence of company logos and brand marks within photos, a task much more complex than detecting varroa mites on bees.
Building the app, backend and other software that is required for the project is a trivial task for the team.
Why hasn't this been done before?
The hardware wasn't available and the idea was probably too far-fetched to be taken seriously. We have the hardware and the team has solved similar tasks. Also Björn, the creator of this project, has a record of finding new solutions. Please visit his beekeeping site. Though in swedish, if you are a Beekeeper you will be blown away and get the idea of extracting honey from frameless supers.
In the initial phase the training of the neuron net is as descibed in the image above, the only difference is that a human has to cut bees with varroa on, and for that we are building a superfast tool. The image just describes how the neuron net works but then again there is a layer above with statistical calculations concluding and considering the overall.
Timing and team
During coming season we will extend our own survey, collect images from around the world and use all this data to produce the software and the app. Initial data are already at hand from our pilotstudy in 2016 and the infrastructure and programmers is developing now.
Our team represents leading experts and advisors from top Universities in Sweden and the U.S. The project has been presented to and collaborates with beekeeping researcher from all over the world.
For the initial stage we ask for just 5 000 USD. Funds are needed for building the database stucture with an API that can handle lots of data from the worlds beekeepers and communicate with the users.
The major assets that are needed for the project, 350 000 USD, we think will be funded late May by the European Innovation Program. We also have investors contributing not to mention a huge nonprofit and volunteering effort from our innovation group and beekeeping friends.
The Kickstarter campaign is a Kickstarter. That is to say both a means for funds but also an important way to engage worlds beekeepers and market the concept.
The more funds we can raise from you, the backers, the faster we can develop and the less we depend on public financing. With proper funding we will be able to push the project ahead this coming season. We are already planning for analyzing video footage and using augmented reality. The technology will enable enhanced accuracy and studying other aspects of bees health and behavior. Our research and development also has the potential to expand in to other fields of monitoring biological systems.
Depending on your contribution we wish to offer:
- Personal analysis of colonies.
- T-shirts with the app icon “BEES CAN - BEE SCAN”.
- Your name on our "Wall of Fame".
- Can advertise in the app and on the support page. -
- Use our logo in your own marketing.
- Your name as supporter on our site.
- Medium donors will get invitations to visit apiaries and meet our research team.
- Major donors: Follow our research on site. We acccomodate you for two days in Sweden forests. FOUNDERS CLUB // Private dinner with the team + your name, company logo, and website link will appear as founding supporters on our website & app. You may bring up to 2 additional guests.
Music tracks in the video:
Composed and performed for this project by Björn.
Brewing Up a New
Risks and challenges
Risks are about resources, money and competence. We don´t have all the money yet that will be required, but we do have the knowledge. Our task is hard says our advising professors, but doable. It will take the best developers to accomplish.
We are budgeting 350 000 USD. Applying for funds from the European Innovation Program. For next steps 2018-2019 we are also looking for funding from the National Honey Program in Sweden. Mainly for information campaigns, marketing and pursuing additional development with new technology using video and augmented reality.
The Kickstarter funding is a starter for the collection and collaboration with worlds beekeepers and is necessary for bringing big data for training our software.
We already have an outstanding team on board and we network with world leading experts in the field. Our programmers are specialists in FRCNN (Fast Regional Convolutional Neural Networks) and confident that their ongoing efforts, in building new tools will solve this.
Please browse our site, www.beescanning.com/eng
- (34 days)