Wild Boar Alert project for visually impaired persons

Hello iNaturalist community.
While searching open source community for a backend for the project I will describe later I stumbled upon https://www.ontario.ca/page/reporting-wild-pigs-ontario service which uses iNaturalist platform.
I am new here, have to check the whole system, the app, but from the first read it seems to be a good platform to host a following project:
Wild Boar Alert for visually impaired persons
In my city (Olsztyn, Poland) wild boars are a real problem. They visit the whole city and because of ASF regulations they cannot be put in the truck and be driven back into the woods, where they belong. Wild boars are coming here in herds and are very dangerous for visually impaired persons using assisting dogs or walking with cane. Such case was described in local press (articles are in Polish, but automatic translation is good):
https://translate.google.com/translate?sl=pl&tl=en&u=http%3A%2F%2Fgazetaolsztynska.pl%2F635913%2CGdybym-szla-z-laska-weszlabym-w-dziki.html
https://translate.google.com/translate?sl=pl&tl=en&u=http%3A%2F%2Fgazetaolsztynska.pl%2F637071%2CBoi-sie-dzikow-bo-ich-nie-widzi-Pani-Kasi-pomoga-teraz-wolontariusze.html
After reading those articles I thought about using a service of some sort (open source is the best) to create a project which will:

  • allow people to report sightings with media attachements (no spammers)
  • have a community who can swiftly verify the reports (Quality Data is important)
  • have an API, which can be used to send Alerts to visually impaired subscribers, WCAG compatible

What do you, iNaturalist users, think about it? Are there any potential volunteers in the community to help out? You would not need to be living in my city to help identify the boars in the reports.
I will try to create a project similar to the Ontario one in few days and experiment with real data - observations of boars by myself, check the accessibility of the service, maybe experiment with the API calls.
Will report my findings in this thread.
Thanks in advance for your feedback.

7 Likes

That’s interesting @poolsztynsku. I know very little about this issue but I do wonder how you envisage a visually impaired person might use these alerts. For example, let’s imagine that you’re able to achieve a fairly high level of reporting of wild boars in Olsztyn (right now, the only reports are some distance away).

Would you be looking to alert a visually impaired subscriber based on new observations being uploaded to iNat? This might generate something like “Alert: Wild boar reported at Nikielkowo, 1.2km from your location, 27 minutes ago (27.08.2020 16:45)

Or would you be looking to alert a visually impaired subscriber to the general prevalence of wild boars in an area that they had traveled to? “Alert: iNat shows 6 nearby wild boar reports in past 6 months, between 0.8km and 5.2km from your location. The most recent was 62 days ago at Nikielkowo, 1.2km from your location.

One of the articles you linked to estimates there are 500 wild boar around Olsztyn. While the females and young will likely be in large groups, many of the males will be solitary.

Some googling suggests that boars have an average total territory of 280–400 hectares depending on the season (there are other data to suggest that migratory European boars have much larger ranges). The daily range appears to average about 30 hectares, roughly equivalent to a 600m-diameter circle.

It seems to me that to be useful for a visually impaired person, you would need to get several fairly precise location fixes each day for a significant portion of these animals. So you might need as many as 100 observations a day around Olsztyn. At lower rates, your visually impaired users would primarily get alerts for boars that are far away, or no longer in the area, and when they did encounter boars, most likely there would have been no relevant observations to warn them.

Overall I wonder whether it’s feasible to build a practical warning system for a widespread and mobile animal such as this.

Also, while it would be useful for iNat to have more observations of wild boar in Europe and may be an interesting subject for some observers, it doesn’t seem like the identification aspect of iNat would be all that relevant here. I’m guessing that most people who are able to capture a photo of a wild boar are pretty confident what they saw!

8 Likes

Maybe if they were submitting possible evidence of boars (footprints, feces, digging) rather than a photo of a boar itself?

4 Likes

Sounds more like a Waze app rather than iNat. You want easy reporting and are only interested in recent sightings. Old data is useful for many tasks but not day-to-day travel alerts.

Thank you for comment and very nice ideas.
At first, I was thinking about simple table view of iNaturalist sightings. I checked out IPhone and Android apps today and it seems that table view is only available on IPhone. Android devices have map and picture grid options only. Also I have found that table view is restricted to general location, not address form of lat lon coordinates. This would result in need for custom data view table. Fetch the data via API, send lan lon coordinates to nominatim like service and display boar sightings with address data.
Journey planner is a very nice idea. Maybe at next step of the project using open-source routing software, which could check nearby sightings data after generating the route.

Estimates are just estimates. I really doubt there are 100 observations daily in my city.

Yes, this was a point to consider from the beginning. How to monitor movement of mobile animals without GPS attached. It is quite impossible to do it in realtime, but I hope that swift information will be valuable. Also more observations will allow to generate hotspotsareas, where a boar sighting is more probable.

This is correct, but in this project identification process would be used as a filter and verification step. The idea is that after successful identification all this alert system would be automatic and without AI involved only humans can be trusted that the boar pictures are really boar pictures. People who would use this alert system can’t verify it by themselves.

1 Like

As @rupertclayton mentioned boars are mobile animals, so having sightings based on evidence of boars (footprints, feces, digging) are good as such for database records (hotspots), but it is too late for a warning project.
It seems that when boars are checking out the biotrashcans, they spend a quite a lot of time in the area.

1 Like

Seems Waze is more like “what’s happening now” application (I do not know the app). I think sightings reports of boars in the city area could be a valuable data when they are not needed anymore (too old to present to the subscribers of the service).

1 Like

Let’s check how visually impaired persons can accesss iNaturalist latest wild boars in Olsztyn data:
Part 1. List view on the main website.
It generates table with observations. This table can be read out with screenreader software.
Some findings (new session in the browser, user not logged in) - if you find an error, please comment:

  • Location is problematic. Olsztyn City, WN, PL (place_id=18225) - all city area is not covered. Olsztyn, WN, PL (place_id=18224) - more than city area is covered. New Olsztyn City area definition is required - compare with https://www.openstreetmap.org/relation/2720007 or https://goo.gl/maps/Zv89kwZtitjbHPHA7.
  • By default Map view is displayed first. User has to change the view to List view. The view parameter subview=table is embedded in the link, so it is no problem - just add to Bookmarks.
  • Records have to be sorted by Observed column, as by default they are sorted by Added column. The sort parameter order_by=observed_on is embedded in the link, so it is no problem - just add to Bookmarks.
  • Place column/observation page doesn’t show full address data - it is a big problem, as pointer data in this column is too general. There is no way to know the street level of the pointer as text data, which can be read out by screenreader. Even when you open the observation page, go to Details, which is a pop-up, you will find coordinates only. Encompassing Places in the pop-up ends at county level. As a helper I write full address of the observation in Notes section. Unfortunately to access this data one have to view each observation page and go to h3 Notes section of the page. A hotfix is to use View on Google button from Details pop-up. It will open Google Map page with coordinates and on the left panel, near the pin there are address data available as text. For OpenStreetMap you can find the address by reverse querying Nominatim like service with Lat and Lon, but it is too much work, as Lat and Lon data are in the pop-up. It would be welcomed to have street and housenumber level in Place column and alongside pointer address data on observation page. Also in my opinion Lat and Lon coordinates, Accuracy, Geoprivacy, View on Google, OpenStreetMap do not need a dedicated pop-up and can be rendered straight away under embedded map in div.map_details container. Pop-up could display Encompassing Places data only.
    Note: A Photo details page Associated observations section also has general data in Place field but it has (Google, OSM) links nearby, so when Google link is selected you are redirected to Google Maps page described above.

Thanks for your responses here. To me this idea of determining boar hotspots is the one that has the best chance of iNat data providing useful info to visually impaired people in your city. It seems feasible that an awareness campaign could encourage quite a few people in Olsztyn to report boar sightings, and in time that could help identify areas with low and high numbers of sightings. Of course, areas that iNat users like to visit will get more coverage than those they avoid.

I feel that achieving more specific alerts isn’t going to be realistic. The chance of a visually impaired person being sufficiently close in time and space to benefit directly from a new report added to iNat seems very small.

1 Like

Let’s check how visually impaired persons can access iNaturalist latest wild boars in Olsztyn data:
Part 2. E-mail Subscriptions
After registration on iNaturalist you can subscribe to a newsletter with sightings in your defined location or selected taxon. This is not realtime information rather daily information containing observations published in the last 24h.
Some findings - if you find an error, please comment:

  • Observations in the newsletter are sent as picture grid view
  • Pictures have alt=“Square” attribute
  • Pictures are links to observation page
  • No text observation details to read by screenreader are available (like in table view from the main site), only text probably from species_guess attribute (as read from json output)

Let’s check how visually impaired persons can access iNaturalist latest wild boars in Olsztyn data:
Part 3. Atom Feed
On iNaturalist you can subscribe to an Atom Feed of every observation page. The feed will be updated automatically by the RSS client.
This is the sourcing observation page:
https://www.inaturalist.org/observations?nelat=53.821599520651816&nelng=20.79286008607596&order_by=observed_on&place_id=any&subview=table&swlat=53.70998605055134&swlng=20.32971769105643&taxon_id=42134&quality_grade=research
Simple GeoRSS Atom Feed for observations above:
https://www.inaturalist.org/observations.atom?verifiable=true&page=1&spam=false&nelat=53.821599520651816&nelng=20.79286008607596&place_id=any&swlat=53.70998605055134&swlng=20.32971769105643&taxon_id=42134&quality_grade=research
XML source example of one entry in such feed:

  <entry>
    <id>tag:www.inaturalist.org,2005:Observation/57936594</id>
    <published>2020-08-29T17:01:29Z</published>
    <updated>2020-09-03T15:51:55Z</updated>
    <link rel="alternate" type="text/html" href="https://www.inaturalist.org/observations/57936594"/>
    <title>Wild Boar</title>
    <author>
      <name>poolsztynsku</name>
    </author>
    <content type="html">&lt;p&gt;&lt;img src="https://static.inaturalist.org/photos/92454587/medium.jpg?1598719873" alt="Medium" /&gt; &lt;img src="https://static.inaturalist.org/photos/92455585/medium.jpg?1598720211" alt="Medium" /&gt; &lt;img src="https://static.inaturalist.org/photos/92456260/medium.jpg?1598720453" alt="Medium" /&gt;&lt;/p&gt;&lt;p&gt;Stado dzików na ulicy Reymonta 26, Olsztyn. Na zdjęciach dorosłe osobniki oraz młode.&lt;/p&gt;</content>
    <georss:point>53.7891614401 20.4963025767</georss:point>
  </entry>

Some findings (feed subscribed in www.newsblur.com app/Feedbro browser add-on) - if you find an error, please comment:

  • Atom button is not easy to find on the source observation page. It is available in the Filters pop-up. Also Feedbro’s Find Feeds in Current Tab option is not working on iNaturalist feeds.
  • Pictures have alt="Medium" attribute
  • Only Added date (<published> tag) and Notes section (<content> tag) is displayed where, as a helper, I write full address of the observation. No other text observation details to read by screenreader are available (data like in table view from the main site). You have to go to the observation page.
  • Lat and Lon coordinates are available only in <georss:point> tag, which in NewsBlur is not displayed. It would be welcomed to have Place data available with street and housenumber level as text.

Some of my phone apps, such as Accuweather, send me notifications, e.g. heat advisory, extreme fire danger, thunderstorms. I hear the notification sound and I know to check it. But weather systems take some time to form, which is why forecasting is possible. It seems like wild boar alerts would need to be in real time, more like an Amber Alert, because a wild boar moves from place to place more quickly than a thunderstorm. How would you coordinate this with your proposed app?

I discovered that iNaturalist is great community and many people are specialized in specific taxa observations. My archive Sus scrofa observations have been promptly identified by others without promoting them as alerts dedicated for visually impaired persons. Soon after identification the status data is available in all described above iNaturalist platform channels.
I think that having dedicated users worldwide, maybe many of them already volunteers is such programs like https://www.bemyeyes.com/, could allow to achieve low observation to identification ratio. I noticed that wild boars tend to spend a quite a lot of time in the biotrash or green fields area, so as having real-time position of a herd without attaching GPS is impossible, having an information that the animals have been seen in the area at least 0:30 minutes or 1h ago is important.
Of course some tweaks in those iNaturalist channels could be needed, but more text data available on the www site, Atom feeds, applications can benefit everyone who is in need of text-to-speech capabilities.
As for coordination - it can work just like bioblitz or the Ontario Wild pigs website. Subscribe Atom feed of selected taxa in selectedplace or boundary box with observations which need research grade and just do what you do in the service, having in mind that in this case time is important and data can serve not only the researchers.

here’s the way i would design your boar alert system, assuming it would be working with iNaturalist as the source of boar sighting data and that users would have a dedicated app as their user interface:

  1. encourage folks to submit data to iNaturalist – even observations without photos – remembering to always identify these as wild boar (as opposed to loading them as unknown and waiting for someone else to identify).
  2. in your dedicated app, have someone plot out the course that they plan to travel (using whichever route planning service you want to use). get the bounding box for that route, and pad it with, say, a 500m buffer.
  3. pull the boar observations (including those without photos) from iNaturalist within that padded bounding box for the last X number of days, then filter those down a little more to prioritize the ones that are within, say, 500m of the route.
  4. reverse geocode those observations so that they can be represented as locations near street addresses rather than as GPS coordinates. a user of your app would be able to review the list at the beginning of their journey.
  5. for those who still have some vision, plot the data as a heatmap layer underneath the route. (for those without vision, if you can provide turn by turn navigation in your app, you could also describe the heatmap levels between each turn.)
  6. you could also have a boar locator function, where you could use the compass and GPS in the phone to allow a person to point the phone in a given direction, and it could provide some cues to indicate whether there are any observations within, say, 500m, within a 5 degree field of view from your location. that would allow someone to use their phone to scan around them to see where the boar observations are spatially.
  7. finally, you could have special account that is subscribed to boar observations within the city. you could check for new observations every once in a while, and as new observations pop up, you could push that information out to your users as a notification of new sightings within the city.

Thank you very much for great and innovative ideas. Unfortunately it seems that step one is the hardest, as iNaturalist is not a popular service in my city (just little above 200 observations in Olsztyn city/county area).

1 Like