Hi! We need more information about this project, in order to make it eligible.
Thanks.
Here is the project I put together to update my Innovation Project Farm Map. It was 6155 photos and as you can see it didn’t work very well, mainly due to an attempt to reduce some of the images and taking the images from a height below what the camera I have is capable of to attempt to obtain a GSD below 1cm. I have uploaded this as an example of what not to do and for you to be able to review how the software works when people like me get it wrong! I have also included some screenshots from the previous attempt (1458 Pictures) so you can see it works this version was processed in 1.2.0 though.
The idea of the project is to reduce the cost of making the images, so I was aiming to complete the site with the lowest image number possible at a GSD and resolution that would make it easy to obtain detailed plant counting or density. What it has enabled us to do is work with the site agronomist as it is easy to identify soil types from the images without further processing. I intend to create it as a heatmap once I can learn how to do it with the eventual aim to export the data to a variable rate spreader for next season. As a small business processing capacity and power is at the lower end until my new server arrives so this project! This dataset was processed on an under powered system.
Wet Gate Farm Pix4d Challenge Orth 6155|690x378
I have put the project log into a dropbox link as people were mentioning difficulties in uploading the large file please let me know if you need it in another format!
Best Regards Matthew McKay
This was a project we undertook earlier in the year, just managing to get it done before the UK’s first lockdown due to Covid.
The site is a power station that is to be decommissioned and the site redeveloped. The primary focus of our survey was to provide a Point Cloud and DSM of the site to enable volume calculations to be undertaken. The main site sites within a flood risk area while the former ash tip to the East side of the site is higher. Our clients are looking at the feasibility of the cut & fill works to use the former tip stock to make up the levels to the main site once the buildings have been removed.
The main chimney was over 200m high and the cooling towers 115m above surface level so we opted for a multi-rotor (DJI P4 Pro) to undertake the work as opposed to a fixed wing Ebee since we’d need to fly between the cooling towers and around the chimney. There was also the issue of National Grid infrastructure on the site that we needed to avoid overflying, therefore a multi-rotor made sense as there is no requirement for long turns at the edge of the survey area.
The project was undertaken over 3 days on site. 1 Day for setting out the 80 GCPs used to calibrate the project to OS Grid with 2 days of flying to capture the data.
The project was originally processed on Pix4D Mapper using a MS Azure virtual machine harnessing Xeon processors, 112GB RAM and Twin Tesla K80 GPUs. It took 7hrs to process the Orthomosaic.
The project has been re-processed on Pix4D Matic as a test project. The results are impressive. The project has been processed on a laptop, Intel I7-8750HProcessor, 32GB RAM and Nvida GTX1060 GPU. The orthomosaic processed in just under 3hrs.
Project details
Average GSD: 1.93505 cm
Project coordinate system: OSGB 1936 / British National Grid - EPSG 27700
Camera name: DJI FC6310
Number of images: 4404
Calibration
- Image Scale: 1/1
- Maximum extracted key points: Automatic
- Reoptimized: true
- Duration: 39m 6s
Completed: 99.7275%
Calibrated images: 4392
Uncalibrated images: 12
Camera optimization: 0.168411%
Median number of matches per camera: 9,151
Ground Control Points (GCPs)
Number of GCPs: 80
GCP coordinate system: OSGB 1936 / British National Grid - EPSG 27700
Units: metre
GCP error [cm]:
mean position error x, mean position error y, mean position error z
0.000979351, 0.000944755, 0.000429536
sigma position error x, sigma position error y, sigma position error z
0.00118959, 0.00111954, 0.000461938
RMS position error x, RMS position error y, RMS position error z
0.00154086, 0.0014649, 0.000630784
Densification
- Density: Optimal
- Minimum matches: 3
- Duration: 7h 5m 46s
Number of points: 542,235,789
Digital Surface Model (DSM)
- Resolution(cm/px): 1 GSD 1.935 cm
- Surface smoothing: 2
- Interpolation: true
- Duration: 1h 11m 15s
Image resolution: 117441x77891 px
Orthomosaic
- Resolution(cm/px): Same as DSM 1.935 cm
- Duration: 2h 58m 52s
Image resolution: 117441x77891 px
Link to log files here:
Bonjour tout le monde,
There is so nice projects on this competition. Here is our project and it’s a special one. The story behind our dataset begin in 2017. It was our very first linear survey with a multirotor drone. The road is now over, so I’m excited to share my project with the Pix4d community. This beautiful dataset was taken just after the road was built in sept. 2020. With the collaboration of government engineers, we flew a last flight over the road to check the road alignment with the CAD design.
Project Details
Location : Sartrouville - Montesson - FRANCE
Site size : 7km linear
Hardware : eBee X equipped with the senseFly SODA3D camera
Data capture : 2 hours
Number of images captured : 1,751
For those who are interested, please check the video I made when the road was under construction :
https://www.youtube.com/watch?v=p7pHjMfP8VY
Regards,
Clément CHAUVET
TEST
You have 5 minutes, Matjaz!
Do you dare to go big?
This topic was automatically closed after 2 minutes. New replies are no longer allowed.
That’s all folks!
The Pix4Dmatic Community Challenge submissions are now closed!
Thank you all for your participation in this Pix4D Challenge! This week we’ll assess the eligibility of the projects and make a live stream of the draw that you’ll be able to follow, live from our social media accounts. More information about this by Wednesday.
If you weren’t able to participate, stay tuned for upcoming challenges here in the community. Better luck time. Let us know in the Pix4D cafe what kind of challenge would you like to participate in.
Hello Everyone!
Don’t miss our live, in 30 minutes, on Instagram and Youtube!
I might draw your name!
Instagram: Pix4D - Measure from Images (@pix4d_official) • Instagram photos and videos
Youtube: http://www.youtube.com/subscription_center?add_user=pix4dsoftware
Facebook: Pix4D
And thanks again for participating!
Congratulations to the winners:
@Biodrone
@lakkuceiiit09
@Matti_Hytola
And thanks a lot to all the 26 participants above!
Looking forward to the next challenge.
Love,
The Pix4D team!