-
Please check the following:
- Is the software that corresponds to the license you obtained installed? The licenses are different for the Pro and Lite versions.
- Are you connected to the Internet?
- Is the license still valid?
- Has the same license key already been used on another PC?
Depending on your network environment (use of an internal proxy), activation may not work properly. There is also an option for offline activation, so please contact us if the problem persists.
-
If you have activated the software online once on a PC, you can use it offline for up to 14 days after the last time you launched the software. After this period has elapsed, you will need to launch the software in an online environment again.
You can also use the offline activation option, which is available even on PCs that cannot connect to the Internet.
Select “Choose another activation method” from the activation screen that appears after launching the software, and follow the instructions on the screen to apply for offline activation.
Please note that if you wish to transfer your license to another PC after performing offline activation once, you will need to apply separately.
-
All of our desktop applications are for Windows PCs only. Basically, you can install them on any PC with Windows 10/11, but if you are going to perform a full-scale analysis, we recommend a gaming PC class specification (RAM memory: 16GB or more, etc.).
A GPU is not required, but installing an NVIDIA GeForce RTX series graphics board can speed up some processing.
-
For the license purchase classification, please check “Terms and Conditions of License Agreement” on the Software Business Page – Purchase Flow. If you have any questions, please feel free to contact us.
Please note that fees will be different if you are using the service outside of Japan, so please contact us first.
-
The tree species identification model we develop is created using training data created based on the recommended measurement specifications below. Therefore, we recommend UAVs that can fly based on our recommended specifications.
Reference) Recommended specifications for drone measurementAlthough it is not required, we recommend a UAV that can measure while changing altitude to follow the terrain, especially in Japan, where the slopes are steep. In addition, if it supports RTK and PPK to improve position accuracy, more reliable analysis will be possible.
-
*The following answers are only for cases within Japan.
In Japan, by using the 10mDEM released by the Geospatial Information Authority of Japan, it is possible to configure flight settings according to the terrain.
RTK-compatible drones such as the Matrice 350 can be flown according to terrain data using the DJI app. The flow will be as follows.
- Convert Geospatial Information Authority of Japan DEM to tif.
This can be converted from DF Scanner’s File＞Import＞Import Geographical Survey Institute DEM. - Adding geoid height
You can check the geoid height of the target area from the site below. Add the geoid height to the tif data by raster calculation using GIS software such as QGIS.
参考) ジオイド高計算 - Loading in DJI Pilot app
There is a section called “Import DSM” in the flight settings, so by loading the data transferred to the microSD, you can set the flight to match the Geospatial Information Authority of Japan DEM.
For devices that do not support RTK, such as the Mavic 2 Pro, you may be able to fly them according to the terrain by using a third-party app.
In a past seminar video, we explain examples of flight design when using camera drones and LiDAR drones.
1st DF Online Seminar
2nd DF Online Seminar - Convert Geospatial Information Authority of Japan DEM to tif.
-
We recommend DF BIRD, developed by our company, and Metashape by Agisoft.
Analysis is possible with other SfM software, but the accuracy may not be stable. In particular, be sure to check whether the software is suitable for compositing forest images.
-
When creating a DSM or DTM, distortion will occur around the edges of the data, so we recommend setting a margin of about 10m from the range of the measurement target.
You can set a buffer of about 10m when setting the range in advance, or if using DJI Pilot2, you can set it when creating a route.
-
When the mobile phone signal connection is good and the target location is within line of sight, the basic approach is to fly using network RTK. This enables highly accurate real-time positioning and reduces the burden of post-processing.
The following measures are taken to prevent position deviation:
– Setting a reference point using a virtual reference point (VRS)
The coordinates of the location where the D-RTK2 is installed are obtained based on VRS information, and after performing coordinate conversion, accurate position information is input to the remote control.
– Implementing post-processing kinematics (PPK) processing
This method performs post-processing analysis based on the GNSS log to obtain more accurate position information.
-
When combining data from different takeoff and landing points, discrepancies in position and height may occur.
When using network RTK, discrepancies are unlikely to occur even if the location is changed, so no special precautions are required.
When installing an RTK station to measure, discrepancies may occur, so separate position correction may be required using point cloud processing software, etc.
In addition, we recommend leaving a margin of about one course when setting the flight range in case of position correction.
-
DF Scanner’s tree species identification model is created using data based on the drone measurement specifications recommended by our company. Therefore, if the drone measurement conditions differ from those recommended by our company, tree species identification may not work properly.
Tree species identification may work well using Site-Tuning identification or My Tree Species AI Identification.
-
Two types of parameters can be set in the advanced settings for Tree top detection.
Smoothing Level
This represents the degree of smoothing of the CHM data. The larger the value, the more the fine undulations in the image will be removed, and only the tops of large trees will be extracted.
Tree Spacing
Set the minimum interval for detecting treetops. Treetops that are closer than the tree spacing setting will not be extracted. It is recommended to use a distance measurement tool to set this based on the distance of the areas where the tree spacing is narrow.
For many forests, the default values (Smoothing Level: 0.8, Tree Spacing: 1.2) are effective, but detection accuracy may be improved by adjusting the parameters depending on tree species and tree density.
- When adjusting the value, we recommend changing it in increments of 0.1 to 0.2.
- The following is an example of a setting in a major coniferous forest in Japan. (Smoothing Level / Tree Spacing)
cryptomeria japonica：0.8 / 1.2
chamaecyparis obtusa：0.8 / 1.2
larix kaempferi：2.0 / 1.2
abies sachalinensis：2.0 / 1.2
-
For CHM calculations, in Japan you can specify the Geospatial Information Authority of Japan data as DTM data, but if you are using it overseas, you will need to prepare the DTM data yourself.
Using a LiDAR drone
DTM data can be easily created by analyzing measured point cloud data with DF LAT.
Using a camera drone
In principle, it is difficult to obtain forest ground information. However, if the image has a large amount of ground, it may be possible to create ground data using SfM software.
Measure by yourself
It is possible to analyze using data published on the Internet. However, please note the following points when using the data.
- DF Scanner only supports GeoTIFF format (.tif/.tiff) data. If your data is in a different format, please convert it before use.
- When downloading data, please ensure you review their terms of use and proceed at your own risk.
- Data published on the Internet is often of low resolution. Analysis with DF Scanner is possible, but the accuracy may not be sufficient.
- After CHM calculation, be sure to check the height of the ground part and perform ground correction operations if necessary.
-
By using “Options” > “Conversion Tools” > “Tree Crown → Treetop Conversion”, it is possible to convert polygon-format tree crown data to point-format treetop data.
-
DF Scanner offers “Sugi Hinoki Identification”, “Hokkaido Species Identification”, “Japanese Species Identification”, and “Dead Tree Detection”, allowing you to identify tree species according to your purpose (Japanese tree species identification and dead tree detection are only available in the Pro version).
As of July 2025, the following tree species can be selected for Japanese Species Identification.
Conifer: Pinus densiflora, Picea jezoensis, Larix kaempferi, Cryptomeria japonica, Tsuga sieboldii, Abies sachalinensis, Magnolia obovata, Abies firma
Broadleaf: Carpinus laxiflora, Ostrya japonica, Cinnamomum camphora, Quercus serrata, Tilia japonica, Castanopsis sieboldii, Castanopsis cuspidata, Fagus crenata, Magnolia obovata, Cornus controversa, Quercus crispula, Fraxinus mandshurica
Other: Bamboo, Dead TreesIn addition to the above tree species, DF Scanner also has a function to learn tree species on its own, making it possible to handle a variety of cases.
-
For identifying Sugi Hinoki, DF Scanner can identify them with an accuracy of over 92%.
For other typical Japanese vegetation, the accuracy is around 60%.If the data is taken outside the recommended measurement specifications or at a time other than the recommended time, the tree species identification accuracy may not be achieved. In such cases, you can create your own training data using site tuning to improve the identification accuracy. Identification accuracy may be reduced if the color of the trees is dark due to insufficient light, or if the trees are red due to pollen.
-
When measurements are taken with a LiDAR drone, they can generally be obtained with an accuracy of within 1m.
When using a camera drone and a DSM created with SfM and the Geospatial Information Authority of Japan’s DEM5A, the error is affected by errors in the Geospatial Information Authority of Japan DEM and errors in the drone’s XY directions, resulting in an average error of about 1-3m. If the Geospatial Information Authority of Japan DEM is 10m or if CHM correction cannot be performed properly, the error will be even larger.
-
DBH is estimated from tree height and crown area.
The margin of error for artificial cedar and cypress forests is approximately 15%.Accuracy may be improved by creating a local formula from the results of a field survey.
For broadleaf trees, estimates are similarly made from tree height and crown area, but accuracy requires verification.
-
There is no position correction function. Please try using the georeference function of QGIS or other software.
-
If you have created a DSM and DTM from laser point clouds and calculated the CHM, correction of the CHM is not necessary.
If you use a camera drone and use the Geospatial Information Authority of Japan’s DEM, correction is necessary due to a difference between geoid height and ellipsoidal height.
-
If the correction method is vertical (default), then generally there is no problem with setting only one point.
If vertical correction does not correct the image correctly, try slope correction or distorted correction. In this case, set multiple ground points so that there is no bias across the entire image.When setting the ground points, make sure to select a location where the elevation has not changed due to recent cutting and filling, etc.
-
Masking is a process that excludes areas other than trees from the analysis based on CHM height information. Although it is not necessary, masking the ground and low vegetation areas will prevent unnecessary data from remaining in the analysis results.
By default, areas with a height of 5m or less are masked. You can adjust the masking height in the advanced settings.
-
The treetop detection function is an algorithm that extracts the tops of cone-shaped trees such as conifers. In broad-leaved trees, the treetop may not be a single point due to the direction of the branches or the clumping. In addition, multiple treetops may occur in tree species such as Sakhalin fir, whose branches point upward.
Please use the editing function to manually correct as necessary.
-
Try decreasing the smoothing strength value in the detailed parameters of the treetop detection function. This will make it easier to detect the apexes of small undulations. Try adjusting the tree spacing value as needed.
Depending on the quality of the CHM data used, adjusting the parameters may not improve the situation.
-
Data acquired with the DJI Zenmuse L2 has a low number of reflection points on the surface of tree canopy, which is known to make holes in the DSM.
If the default settings do not work properly, try changing the resolution when creating the DSM to 10cm or 15cm.
-
DF LAT Pro version v1.2.0 now supports class classification. When detecting the ground, it is possible to assign classes to parts other than the ground and output them. Please refer to the software manual for details.
Note that if you detect the ground with the DF LAT Lite version, you cannot assign classes and only the point cloud of the ground part will be output.
-
As of June 2025, we have confirmed that DF BIRD 1.3.0 or lower versions have a problem where the software crashes when performing analysis on a PC equipped with a GeForce RTX 5000 series GPU. This issue is scheduled to be fixed in the July 2025 update.
-
DF Walker can be used on smartphones and tablet devices, but it is easier to use on a tablet with a larger screen. Also, if you intend to use it for serious drawing operations, we recommend a device with a pen.
- For the Android version, we recommend the Galaxy Tab S series.
- For iOS, we recommend using an iPad and Pencil.
When using location information, please make sure that your device has a valid GPS sensor and geomagnetic sensor (direction). Please note that we cannot guarantee that this app will work on all devices.
*We have received reports that orthoimages do not display properly on some older Android devices, and this has not yet been resolved.
-
On some Android devices, when displaying orthoimages, the image part becomes completely black. This may occur especially on older models or with older versions of the Android OS.
As of July 2025, there is no prospect of resolving this issue. If possible, please try updating the OS, and if that does not resolve the issue, please consider using a different device.
-
When attempting to open a large file, the app may be forced to close due to insufficient memory. For polygon files, you may be able to display them in DF Walker by reducing the file size by clipping it to a few areas or deleting unnecessary attribute information.
In particular, polygon files after DF Scanner analysis tend to be large in size due to the large amount of information. As of July 2025, we are currently considering optimizing the app’s performance.
-
This is due to the image resizing process implemented to maintain the performance of the app. When uploading GeoTIFF data such as orthoimages to the server, if the file size is larger than a certain size, a conversion process to lower the resolution is automatically performed.
To maintain the resolution, be sure to split or clip the image to reduce its size before uploading it to the server.
