311
Deep learning CV and Earth Observation data. What satellites can see? Monitoring of Environment, Land management; Urban planning and Mapping, Mapflow.ai etc. Сontact us at hello@geoalert.io or @godnik0
#UrbanMapping – The Tashkent region construction areas dataset has been updated (github link). This detection was conducted using satellite imagery basemap and is mostly relevant to summer and autumn 2023.
Some extracts:
- The number of detected constructions has almost not changed compared to the year 2021 (around 500 objects covering the agglomeration area of 1000 sq.km)
- The biggest area, given 2023 layer is the construction of the new Uzbekistan's Olympic City which started in 2022.
We hope that there are many more insights that are meaningful for urban researchers and authorities.
Of course, the results are produced fully automatically and may contain errors (there are false positives, construction site contours are not as accurate as might be required...). But the most important that you can use and edit this data, and if you like to contribute back – write an issue and propose the updated vector data that was validated.
If you are interested in the more recent data and the data updates or validation for your favorite city – you are welcome to contact us.
Geoalert RnD #Team is growing. 🚀
Doniyor has joined us (UZ office) as a software engineer. Welcome on board! 🤚
#Mapflow #lab
Experimenting with AI mapping of rooftops.
Here we combined "Buildings" and "SAM Aero" ("segment anything") to get more details on the rooftops using aerial imagery of Australian residential area.
If you think it might be useful for your project you can use this or another custom AI mapping - contact us to join Mapflow mapping lab.
#Buildingfootprints + Vegetation = #Mapflow #custom
Here we tested Bing aerial imagery for some areas in Australia to map buildings and trees in a one-shot processing. The joined output has a valid topology and the mapping class is indicated as the "object type". Behind these experiments is a user story to produce 3D animations for infrastructure development.
BTW we seek more cases in which Mapflow models can be applied in different combinations. 😊
#Mapflow #How-to #Hint
How to use Nearmap as an imagery provider with Mapflow Custom URL.
Nearmap provides access to its Vertical and Panorama Imagery via a Tile API. If you have a subscription to their service you can use it easily with Mapflow Custom URL.
The URL format should be:
https://api.nearmap.com/tiles/v3/{tileResourceType}/{z}/{x}/{y}.{format}?apikey={YOUR_API_KEY}
Playing around with Q2threeJS for #powerline vegetation reports.
Read more about the solution for analysis of vegetation risks and work control near powerlines.
#QGIS #Mapflow #Solutions
A special combined scenario can be created to apply multiple models to the same image – to avoid many uploads/downloads and optimize processing time. The Geojson output contains the topology valid geometries with the attributes.
In this preview two scenarios were combined: Buildings with heights (aka 3D #Buildingfootprints) + Vegetation with height classes (low, mid, high).
You can find more about #Mapflow pipelines in our docs.
Presenting about Mapflow user feedback at the friendly coworking place in Istanbul #Team #Geoalert
Читать полностью…
#GeoQuiz Guess the city.
The right answer will come in the next posts. And the dataset as usually will be availalbe in the urban #mapping open repo.
Mapflow-QGIS Plugin 2.6.0 released!
A neat feature that was always presented in #Mapflow API is now promoted to plugin - you can now create and manage projects, which is useful to organize your processings when it comes to many items.
For every processing you can now download AOI to re-use it in your further research and visualization.
And you can use the "magnifying glass" button on the left menu to preview every Data source that is selected for the processing.
Another #research work featuring #Mapflow plugin for #QGIS at the stage of #buildingfootprints collection. This time a group of researchers from the Institute of #Renewable Energy leverages a machine learning approach for the analysis of PV installations in urban areas. The authors come up with the demonstration of ML technology as a cost-effective way to obtain data for decision-making in the fields of energy and urban planning.
While in this work the building footprints were combined with the DSM data available for the study area, it's worth noticing that Mapflow is capable of generating 3D building footprints (by predicting building heights with ±3m RMSE using mono images only).
One of the most interesting applications the research work refers to is to find out the cost-effective approach to define the best possible location for the solar panel installations given the rooftop area, its insolation capacity, and the other factors that are significant for estimation of the overall potential in the specific urban area.
Check out the publication. 👈
This sort of challenge. A lot of false negatives vs. many false positives. Trying to find a balance.
#Mapflow #Buildingfootprints #Aerial #Africa
At the beginning of the year, we looked back to highlight some milestones in our company life, business, and product development.
We’re thankful to our customers, partners, and users for their inspiration, feedback, and support.
Read the story...
Close to the end of this year Andrey and Rustam joined out team in #Tashkent 👏
Читать полностью…
Upcoming releases – new building model
We are preparing the release of the new building segmentation model. The work is in the active testing phase – using testing areas to calculate metrics and visually assess the new model performance on different construction patterns across the Globe.
Join our beta-testing in progress for early access to the new model and rate the results with the #Mapflow rating tool.
We'd appreciate your feedback! You only need to send us your email which is registered for the Mapflow account.
https://geoalert.io/?subject=Beta#contacts
#GeoQuiz
The unexpected application of Segment-Anything (#SAM, Zoom 18) using Mapflow.ai.
Try to guess what, where? 🤔
Users who send us the correct answers will receive a promocode for 500 Mapflow credits. 😉
Some users keep experimenting with SAM for field boundary delineation. And they even get relevant results (even using an input image of not a very high quality). 😎 Don't hesitate to leave your feedback like this user as we are collecting more evidence to tune SAM for specific object classes. Mapflow powerful capability to run analysis on a large scale + tuned SAM can become a game changer for your mapping tasks. ✨ #India #fields #SAM #Mapflow
Читать полностью…
#Mapflow 3D #building footprints generated from 2D 0.5m RGB satellite image.
錦田, Luoyang, 泉州台商投資區, Quanzhou, Fujian, China
В эти дни проходит конференция «Современные геоинформационные технологии 2024», организатором которой выступает компания Дата Ист.
Коллеги продемонстрировали интеграцию с #Mapflow AI, которая предоставляет возможности обработки космических и аэрофотоснимков совместно с инструментами интерактивной карты платформы CoGIS.
Вкратце, как это работает:
1) Создание нового пользователя Mapflow (необходимо получить API токен)
2) Создание новой обработки
3) Получение результатов от сервера Mapflow и дальнейшая работа с ними в CoGIS.
✍️
Демонстрационный прототип интегрированного решения Mapflow <> CoGIS.
Документация Mapflow #API.
Обращайтесь к нам на hello@geoalert.io, если у вас появились вопросы по разработке и внедрению решений на основе AI картографии.
Geoalert's workshop at the Mapacon Online conference (in Russian) 👆
Читать полностью…
We got to share a new customer success story by Island Green Power whose mission is to help countries increase their solar energy usage, making more renewable energy whilst drastically reducing carbon emissions. So they design solar plants and map the projects.
"Mapflow Qgis integration assists us in swiftly identifying hedgerows and forests during the preliminary analysis of environmental and landscape constraints. Thanks to Mapflow, we have significantly streamlined the preliminary design processes for solar and BESS projects." - says Antonio Saponaro, Project Designer.
Thank you for sharing your story 🤗
This week our #team is in Istanbul for the teambuilding event ❤️. There are nine of us, not all could come. Istanbul is beautiful as always even through the bloody and dark times in history.
On Friday, 10.30 am – 1.30 pm, we will hold a seminar about Mapflow AI for our local partners (we will talk about some insightful user feedback and present the new experimental workflow to analyse imagery with the text prompts). Let's meet in real life by this chance. Feel free to join, we'll send the details about the event - register here https://lnkd.in/e6-cpbEt
Fields model deprecation 👎
🚜 Fields (hi-res) - agricultural fields boundaries detection model - has been deprecated. It didn't meet our standards and we decided to remove it from our default models list in #Mapflow, so we could focus more on our best performing models and get back to this one as soon as we have more resources.
From now on users will not be able to create new and copy old processings using this model, however old results will be kept and available.
If you want to keep using this model (in some territories it performs good enough and was used for research and commercial purposes) - drop us a message and we will connect it to your Mapflow account.
Sometimes you are supposed to be AI chatting with the visitors. Otherwise they don't want to continue talking to you. 😅
BTW, if anyone is experienced in developing #ArcGIS plugins and eager to integrate #Mapflow API - we are ready to discuss and support you.
Mapflow user #feedback. #Vegetation in Urban landscape. Montreal, Canada.
Читать полностью…
Exhibition of the maps and cartographic design at the online conference #Mapacon organized by @cartetika_channel 🔥
Читать полностью…
Looking back at the past year 2022 we decided to write about Mapflow users' feedback ⭐️
To tell the truth we underestimated how insightful this feedback could be. We commit not to leave the free users’ feedback unresponded and will do our best to suggest the solution in case of the issues.
Take a look into our Mapflow feedback research.
Street textures in North #India, Haldwani. #Roads model, Mapflow.
User #feedback
Short #GeoQuiz
What kind of place is in this satellite image?