top of page

In construction and civil engineering sites, reverse staking plays an important role. Generally used to mean "positioning" where stakes or marks are placed on site according to the coordinates on design drawings, in practice it is not uncommon for surveys and as-built control to be carried out all at once at the end of construction. This reverse-staking–like approach of "measuring everything after construction" under limited personnel and time is often inefficient and carries the risk of rework if mistakes are found. This article focuses on photogrammetry, a photographic measurement technology—Photogrammetry (SfM)—and explains how conventional reverse-staking work is changing and the latest methods that simplify 3D as-built recording.


What reverse staking means: definition and background on construction sites

Originally, "reverse staking" refers to the work of accurately restoring construction positions on site based on reference points and design coordinates, i.e., the positioning where surveyors use instruments like total stations to lay out stake points. However, due to site conditions and labor shortages, the term "reverse staking" is sometimes colloquially used for the practice of measuring everything at the end of construction to reconcile the records. In other words, construction proceeds first, and the survey is done retroactively at the end to verify and record the as-built conditions.


Traditionally, for such late-stage as-built measurements, experienced engineers compare drawings with the site and check whether the work has been completed to the design dimensions at various locations. They measure pavement layers and structure widths and heights, check deviations from centerlines, and the series of tasks require much effort and care. If deviations from design values are discovered in the final measurements, corrective work or consultations are required, so sites tend to adopt a mindset of "as quickly and accurately as possible," yet in reality they often end up measuring everything at once at the end. This reverse-staking style hides issues related to workload and ensuring accuracy.


Challenges of conventional as-built control methods (labor reduction, accuracy, record creation)

Conventional as-built control involved a process of manual measurement and recording at each survey point. For example, in road construction, transverse surveys are taken at regular intervals, and widths, thicknesses, and heights are measured with tapes or total stations and recorded. This method has several challenges.


Labor and time burden: Surveying usually requires two or more personnel, and the more measurement points there are, the more days are needed. Heavy equipment must be set up and moved repeatedly, so the work was difficult to streamline.

Limits of measurement coverage: Measuring by cross sections cannot capture locations other than the measured sections. The site cannot be understood in a surface-wise manner, and local irregularities or errors between sections may be overlooked.

Accuracy and human error: Manual reading of angles and distances and transcription of measurements carry the risk of human mistakes. With fewer skilled personnel, ensuring consistent accuracy regardless of who measures has become a challenge.

Time-consuming record creation: As-built control requires taking photos with blackboards and transcribing measurement results into specified forms, so document preparation also takes time. Compiling records from many survey points and creating charts for inspection submission is complicated and a major burden for site supervisors.


As described above, attempting to measure as-built conditions at the end with conventional methods requires processing large volumes of data while ensuring accuracy under labor and time constraints. Therefore, more efficient and comprehensive measurement methods have been in demand.


Principles of Photogrammetry (SfM) and its evolution as a surveying technology

Photogrammetry, a technology that reconstructs 3D shapes from photographs, has recently brought major advances to surveying. In particular, the spread of the method called SfM (Structure from Motion) allows anyone to generate high-accuracy 3D models from multiple photos taken with a camera.


The principle of photogrammetry is simple. Take overlapping photos of the target from various angles, and the computer automatically matches feature points between photos. SfM algorithms simultaneously estimate each photo’s camera position and orientation and compute the spatial coordinates of the feature points, producing a dense point cloud dataset. Further MVS (Multi-View Stereo) processing can densify the point cloud and generate a smooth 3D mesh. In short, by analyzing photos taken with a regular digital camera or smartphone, you can convert the site’s terrain and structures into comprehensive 3D data.


Where photogrammetry once required specialized equipment and craftsperson skills, SfM technology has transformed it into a digital surveying tool anyone can use. Applications range from creating terrain models from drone aerial photos to measuring dimensions of small structures from smartphone images. The Ministry of Land, Infrastructure, Transport and Tourism is promoting the application of 3D measurement technologies to as-built control as part of "i-Construction," and guidelines for UAV photogrammetry and terrestrial laser scanner–based as-built control (draft) are being developed. Photogrammetry is particularly noteworthy for enabling high-density measurement with inexpensive equipment, and it is expected to contribute to improved productivity on sites.


Affinity of "reverse staking" × SfM: recording surfaces with few people and obtaining high-accuracy overall understanding

The combination of late-stage "reverse staking" style as-built control and photogrammetry (SfM) is highly compatible. It allows a small team to cover wide areas and record as-built conditions in a surface-wise manner.


Using photogrammetry, for example, one to two people can walk the site taking photographs and obtain 3D data for the entire construction area. There is no need to carry heavy tripods and repeatedly reposition them; you can move nimbly with a camera in hand even in confined sites or terrain with elevation differences. If drone flights are possible, the entire site can be captured from above in a short time, which also offers safety advantages.


The resulting point cloud model contains information for every location of the constructed elements. Rather than judging from a few cross sections as before, you can visualize the as-built conditions everywhere, making small deviations from the design less likely to be missed. Moreover, because arbitrary dimensions can be measured from the data later, you won’t have to worry on site that "we forgot to measure that point."


In terms of accuracy, if photogrammetric measurements are performed properly, the overall shape can be reproduced with errors within a few cm (within a few in). By placing several ground control points (points with known coordinates) you can improve the model’s overall coordinate accuracy. If photos are taken with an RTK-GNSS–enabled drone, high-precision position information is embedded in the images, so the 3D reconstruction can be achieved with cm-level accuracy (half-inch accuracy) even with few control points. In other words, photogrammetry can be an ideal solution to reduce personnel in reverse-staking work while maintaining accuracy.


Shooting tips and 3D processing workflow (using smartphones, digital cameras, drones; control points; cloud processing)

Here are the procedures and tips for carrying out as-built measurements using photogrammetry. Use equipment such as a smartphone, digital camera, or drone depending on the situation, and pay attention to the following points.


Equipment selection and shooting plan: Decide the shooting method based on site scale and the target. For narrow areas or detailed recording of structures, use a high-quality smartphone or digital camera for close-up shots; for large earthworks or hillside surveys, drone aerial photography is effective. Plan the shooting area and route in advance and ensure photos overlap to cover the entire area.

Sufficient overlap in photos: Photo overlap is important for SfM analysis. Approximately 70% or more overlap between adjacent photos is ideal, and taking photos of the same area from various angles improves accuracy. Ensure consistent focus across the subject and avoid blur or overexposure. For drones, automatic flight can take images at equal intervals; for hand-held shooting, step over gradually to shift positions between shots.

Placement of control points: To scale and georeference the model, include known coordinate points in several photos if possible. Place ground markers (target sheets or spray marks) and later survey their positions with GNSS or a total station. Setting these control points in the software allows you to align the output model to a geodetic coordinate system or improve absolute accuracy.

Cloud processing for analysis: After shooting, import the photos into photogrammetry software or a cloud service. Nowadays, cloud services that generate point clouds and output models simply by uploading images are well established. Even without a high-performance PC, cloud processing can deliver results quickly. Depending on the number of photos, processing at the scale of several hundred images often completes within a few hours.

Checking results and preparing for use: Review the output 3D point cloud or mesh model for gaps or distortions. Remove noise points or correct offsets from reference surfaces as needed. If control points were used, check the errors to ensure the model aligns with the designated coordinates. Once ready, proceed to the next utilization steps, such as importing the model into CAD software or creating longitudinal and cross-section drawings.


With the above workflow, you can obtain a 3D as-built model even with just a smartphone and without special shooting equipment or advanced skills. What matters is sufficient photo coverage and appropriate referencing: with site ingenuity, high-accuracy measurement is achievable.


Uses of the generated 3D model (cross-section comparison, volume calculation, report creation)

3D models (point cloud data) obtained by photogrammetry can be used in many ways for as-built control. Here are some analysis and documentation methods that were not possible with conventional planar records.


Comparing cross-section shapes: Extract longitudinal or cross sections at arbitrary locations from the point cloud and overlay them with design cross sections for comparison. For example, you can visually evaluate whether a slope gradient is as designed or whether embankment shapes fit within prescribed cross-section shapes. If you display differences between measured and design surfaces on 3D data as a colored map (heatmap), areas of excess or deficiency in height become immediately apparent and help objectively determine pass/fail.

Calculating fill and excavation volumes: By comparing the completed terrain point cloud with pre-construction or design surface data, as-built volume calculations become easy. Instead of calculating soil quantities by cross section as before, you can automatically compute the volume difference between models and quickly grasp actual quantities of fill or excavation. This is powerful for progress control and verification of quantities, and it streamlines the preparation of reports to clients.

Efficient creation of forms and drawings: You can read required dimensions from the 3D point cloud to support or automatically generate as-built control charts and inspection forms. For example, measure heights at prescribed locations in the model and compile differences from standard values in a table, or annotate measured dimensions on orthoimages created from point clouds for submission drawings. Photogrammetric data can be preserved as a digital photo ledger or 3D construction record, making it valuable for later verification or maintenance documentation. Sharing the 3D model among stakeholders allows intuitive explanations of as-built conditions and provides persuasive evidence during inspections.


By utilizing the acquired point cloud data in these ways, many previously manual measurement and drafting tasks can be semi-automated and accelerated. Site supervisors and surveyors can reduce the burden of repetitive tasks and focus more on analyzing data to improve quality.


Success cases and expected scenarios: confined sites, urban redevelopment, temporary structures, etc.

Here are some scenarios where integrating photogrammetry into reverse-staking surveys proved effective.


Construction sites in confined spaces: In urban sites, finding a place to set up surveying equipment or securing line-of-sight is difficult. In one sewer project, a single worker hovered a small drone above a trench just wide enough for the roadway and photographed the entire site within minutes. From the resulting point cloud model they confirmed pavement restoration thickness and gutter slopes, and as-built measurement that used to take half a day was completed in under an hour. Photogrammetry’s ability to flexibly shoot from above or around confined sites demonstrates its mobility and enables fast, accurate as-built control.

Urban redevelopment (top-down construction method): In the reverse-staking method used in top-down construction for building basements while constructing the upper structure, it is necessary to check for column position deviations and accumulated floor height errors routinely. At a redevelopment site, internal spaces were recorded by photogrammetry after each floor’s frame was constructed, and displacements of temporary columns and earth-retaining structures were measured on 3D models. Even in dark areas, illumination allowed high-density point clouds to be acquired. By comparing with design BIM data, column positions were confirmed to be within tolerance and corrective feedback to the next processes was delivered quickly. This changed surveying in confined spaces from being reliant on craftsmen’s intuition to data-driven verification, leading to improvements in quality and safety.

Management of temporary structures: Temporary structures such as temporary bridges or shoring installed during construction are often thought to be outside as-built control because they will be removed, but measuring them can be useful for understanding stability and surrounding impacts. At one site, the settlement of a temporary yard for a large crane (area where timber mats were laid) was periodically measured by photogrammetry. By comparing ground point clouds before and after machinery operation, settlement amounts were mapped in a surface-wise manner, aiding early detection of warning signs. For temporary work platforms, 3D capture immediately after assembly recorded component placement and heights; these records served as verification data if problems arose after dismantling. The ability to easily record temporary structures is a unique advantage of photogrammetry.


In these scenarios, photogrammetry-based as-built control contributes to labor reduction and risk reduction. Because shooting methods can be adapted to site conditions, photogrammetry can provide insights that conventional surveying methods could not obtain even in special cases.


Conclusion: smartphone-based simple surveying case studies and future outlook

3D as-built recording using SfM photogrammetry is significantly changing reverse-staking work. As a means to achieve both labor reduction and efficiency while maintaining accuracy and objectivity, it is likely to become standardized at many sites in the future.


Recently, simple surveying systems that combine smartphones with small GNSS receivers have appeared. For example, using solutions like LRTK Phone, a smartphone becomes a high-precision positioning device, allowing anyone to easily perform reference point surveys and photo measurements. There are cases where small and medium contractors have adopted such systems to perform drone aerial photography, generate as-built point clouds in-house, and use them in inspection documents. The convenience of completing the process with a smartphone means young engineers can learn it in a short time, and the ability to start 3D as-built control in-house without relying on external specialists is a major attraction.


The important thing is to try it on site, even in a small area. A 3D model can be created from as few as several dozen photos. Simply finding and comparing the same dimensions you used to measure conventionally can demonstrate photogrammetry’s usefulness. If the data are shared in the cloud, supervisors and clients in remote locations can grasp the situation in real time.


With photogrammetry as an ally to reduce the burden of reverse staking and ensure as-built control, the way people work on construction sites is steadily beginning to change. Record an entire site with a smartphone in hand and proceed to inspection on the spot. That new daily routine is not far off.


Next Steps:
Explore LRTK Products & Workflows

LRTK helps professionals capture absolute coordinates, create georeferenced point clouds, and streamline surveying and construction workflows. Explore the products below, or contact us for a demo, pricing, or implementation support.

LRTK supercharges field accuracy and efficiency

The LRTK series delivers high-precision GNSS positioning for construction, civil engineering, and surveying, enabling significant reductions in work time and major gains in productivity. It makes it easy to handle everything from design surveys and point-cloud scanning to AR, 3D construction, as-built management, and infrastructure inspection.

bottom of page