Inside the app

Why is there a silhouette bee when I want to add an image?

Move your camera to sync it with the size of the bees and your picture will be optimal for analyse.

Where in the hive should I take the pictures?

Take pictures from second brood frame and as many as you want to cover 3000 -6000 bees.

How should I take the picture?

Place the comb so you can hold the phone with both hands. Less shaky. Avoid direct sunlight since it creates sharp shadows and contrasts. Take the picture from about 25 cm from the frame in right angel. Unsharp, to close or far away images will impair results. On a Modified Dadant (448x286 mm) frame take 4 images on each side. On a Medium 3/4 Langstroth (448x159 mm) take 2 images from each side. Avoid overlap.


What is a collection?

A collection contains the images you take from a hive or even an apiary. Results from analysing will reflect the average of the collection.

Can images be stored in different collections and sent for analyse later?

Yes take images and build collections offline, press scan when you are ready and connected.

How do I create new collections?

Tap button in bottom bar.

Can you edit metadata, like apiary, hive etc?

Tap bottom right button with +. Choose "No origin selected". Select origin or edit choosing "Manage my origins". A tip may be to prepare origins before starting the work. Then its easy to just select the specific hive when needed.

Is there a way to search images, collections?

In Gallery view, choose "Filter collections". Select race, hive and/or apiary.

How do I delete an image or a collection?

Long press and options will reveal.

How many images is maximum in a collection?

Maximum 30

How many images should I take from one colony?

Depending on infestation level you will want to picture between 3000 -6000 bees. With 250 bees in each image that is 12 - 24 images, covering 2 - 4 brood frames.

Where in the hive should I take the pictures?

Take pictures from second brood frame and as many you want to cover 3000 -6000 bees.

Why not take pictures from frames above the excluder?

Mites are preferably found on the brood and results tend to vary more the further away you get.

When should I take the pictures?

Preferably in the beginning of the season and just after your main honeyflow, 1-2 months before wintering.

Can I access my images and results?

Yes they can be accessed within the app.

Can I change language?

In update. For now faq is only in English and Swedish.

How to zoom?

Pinch with two fingers.

About the app

What cellphone is needed?

On android 4.1 and on iOS 8.0

Where can I download the app?

Available on Appstore and Google Play

What is the price of using the app?

Free to download. Free to scan up to 30 images in your first collection. You will start with a sign up bonus of 10 Tokens. Analyzing one collection is 10 tokens = 1,99 €. That equals about 0,06 €/image 50 Token Pack - 5% € 9,49 100 Token Pack -10% € 17,99 500 Token Pack -15% € 84,99

When should I take the pictures?

Preferably in the beginning of the season and just after your main honeyflow, 1-2 months before wintering.

Is the app depending on internet connection?

You can take pics offline and send for scan when you have connection. Analysing varroa infestation is done on our servers.

What is varroa infestation?

Varroa infestation is the number of mites on the bees divided by the number of bees. In a sample with 3 mites on 300 bees that would equal 3/300 = 0,01 = 1%. So that is the number of phoretic mites, (mites on bees not in the brood) detected in the images in relation to the number of bees. The detected infestation level in the images is then multiplied with a factor. We are constantly evaluating what factor we use. The factor reflects the ratio between observable and actual number of varroa mites on adult bees as measured with alcohol washing. (The factor may be subject to change as our research progress.) The resulting figure is then divided with the number of bees in the images analysed. Example: 3 mites are detected. 3*factor 8 (arbitrary value)=24 mites calculated. 24 mites per 2500 bees in images = 24/2500 = 0,0096 ≈ 1% infestation.

What is Deformed Wing Virus, DWV?

Varroa delivers viruses when it ingests on its prey. Presence of DWV is a strong indicator of high infestation level, though absence is not a reliable sign varroa is not at hand. Thus you may want to check closer when DWV is detected. The analysing accuracy is not yet very reliable but will improve soon.

Is the app reliable?

On launch we only know that in our lab environment, the software finds the Varroa, the DWV-bees and the Queen with great accuracy. The more images we can collect from you using the app the better it will become. The neural network will learn from its errors and improve. Look at it as you are part of a great pioneering mission. In the first versions prepare for missteps and report them, helping our common progress. Disclaimer: Results are based on our field studies and may depend on our location, methods of manual analyses, breed of bees. Please read more in https://beescanning.com/eng/studie/ Research results from 2017 are preliminary and will be published spring 2018. If you wish to calibrate the image analysing results with your own study washing samples with alcohol we would appreciate  your feedback and incorporate any new knowledge that may be due to your circumstances. Have in mind that any result of low infestation is not a guarantee you don't need to treat. Depending on your individual conditions; as what type of bees, time in season, reinvasion from neighbors, quality and number of images etc may influence the interpretation of the result. On the other hand if a high infestation is recorded that is of course a strong signal to be taken seriously.

How can  results bee used?

Results will indicate if you may need to treat the colony. Results may soon even be used to find trends and colonies with varroa resistance treats. 0% Great! You may want to confirm this result by collecting more images and/or examine with alcohol washing. The colony may be of interest for breeding resistance. <0,5% Depending on time of season this may be ok. Note the colony for further evaluation. 1% Plan for treatment, monitor closely. 3% You will want to act soon! This is a threshold! 5% There is danger! This colony is in jeopardy and may be a source of reinvasion in your neighborhood. Act immediately.

How should I treat the colony?

There are many options. Depending on time of season and your ways with bees you will have to choose your preferred method. Please read more at http://scientificbeekeeping.com/varroa-management/treatments-for-varroa/

Where are scanned images stored?

Images are stored on our servers.

Who owns the images?

You. We own the rights to use them according to your agreement in Terms of Service.

What happens if I loose my phone?

Your images may be recovered from the server.

What does the app icon symbolize?

It radiates the stress bees show from the small crabish parasites. As well as internet communication. The dynamic hexagons signals bees, the red dot varroa and danger, wifi stripes signals communication.

What is the accuracy of the image detection of varroa?

In lab environment the software performance error is about +/- 5 %. That is how accurate the NN reports compared to the known test data. In reality the error may be greater because of many factors. For one we don't really know how mites behave on different kind of bees under specific conditions. The NN can find the mite even if only parts of it are visible or if its under a wing but results may be affected due to optical circumstances. We will keep learning how to integrate uncertain detections of difficult objects.

What is the accuracy of image analysing counting bees?

In lab environment the software performance is about +/- 5 %. That is how accurate the NN reports compared to the known test data. Counting is done by a totally different technology compared with the mite detection. We are rather certain the performance will improve fast.

What is the accuracy of detecting the Queen?

In lab environment the software performance is better than +/- 5 %, it is actually a bit better. Detecting the queen is a much easier task than the varroa.

Can the app find the Queen instantly?

We are planning to enable offline capabilities in updates

What is the accuracy of DWV detection?

In lab environment the software performance is not yet reliable +- 20% but improving. We are adding it now and would be grateful for more images with DWV to train the neural network. The NN seems to falsely identify some new born bees with DWV. Thats understandable because the NN seems to connect features of young bees with DWV. Older bees seldom show DWV since damaged ones normally are quickly eliminated.

I paid but the server was not available

Check you have an internet connection. Please try again later. If problem not solved mail support@beescanning.com We are happy to help in any way any time.

GDPR?

Please mail any question to support@beescanning.com

Project

How does the neural network learn?

The neural network, NN, 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. However this is 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 for the development team was to find the way of training that is optimal for this particular task. Finding very small objects in a huge amount of information.

What if you don't find enough training images for the neural network?

A crowdsourcing campaign secured resources to develop Tagger. Our tool to structure images and manually label regions to train the NN. We have collected about 4000 images and will contnue to use the data gathered from the app as a tool for further ongoing training of the neural network.

Is the development team actually capable of solving this task?

Yes! The image analysing 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 as complex 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. Teorem has the hardware and the team has solved similar tasks. The innovator Björn Lagerman has a long record of combining knowledge in new and creative ways as well as the experience to lead complex projects.

Where can I upload images?

Please collect images. Especially from frames where brood is visible and with varroa on bees. Upload on http://tagger.beescanning.com and contribute by labeling regions.

How is the project financed?

Initially we were funded by backers on Kickstarter. Then we received financing from the European Innovation Program, EIP. We are also supported by Almi Örebro and Handelsbanken in Lindesberg. Please read more at http://fribi.se/sponsorer/