The artificial intelligence that writes daily construction reports

March 29, 2023

It is one of the most unpopular activities, but nevertheless a very important one: keeping the daily construction report. It does not require much time, but special attention and care. Neglecting it can have expensive consequences during the warranty period.

The construction diary was one of the first processes to be digitized on the construction site with the introduction of simple apps. But after digitization, the core problem remains: The construction diary still has to be kept manually and is therefore unpopular and neglected. It's time for a revolution: the self-writing construction diary.

What is the self-writing construction daily report?

Conventional digital construction diaries enable simple and collaborative documentation of the construction process via computer or smartphone. Copying functions or weather data facilitate the daily work. However, the work performed still has to be recorded manually. With the self-writing daily construction report from oculai, the work performed is now also pre-filled automatically. Cameras and artificial intelligence are used to automatically record and document construction processes.

How does a self-writing construction report work?

At oculai, the observer role of a foreman or site manager is taken over by intelligent crane cameras - cameras that are mounted on the crane tower and record the construction site from a bird's eye view. The videos generated are then run through various algorithms that recognize and document construction processes from the pixels in the video. These algorithms are developed by oculai and are based on artificial intelligence (AI). You can think of it as a computer watching the video and taking notes of what work is going on where and when. The output format of the data is at the end for each day a start time, end time, component (e.g. "3rd floor, section 2") and the recognized work (e.g. "concreting ceiling").

At the end, a report is generated for each day. Thereby oculai automatically enters the working hours, the weather and the recorded activities and components. If desired, oculai automatically adds before and after pictures. Information about personnel deployment or even deliveries and machines can easily be taken over from the previous day. A calendar overview also shows bad weather events and special occurrences. The daily construction report can be extensively configured and exported with individual content. Click here for a sample export of an AI-generated construction day report!

The construction diary in the calendar view
The overview as a calendar with automatic bad weather detection and messages

What operations are covered by oculai?

Currently (March 2023), oculai distinguishes between 36 different operations that can take place outside. These include operations such as (semi) precast, masonry, various in-situ concrete works or waterproofing. Since oculai is typically used on cranes for shell construction, the AI is limited to work that primarily takes place in the shell and is recorded by the cameras from above. However, oculai is not only used in shell construction, but also on infrastructure projects where top-slewing cranes are used. However, the digital construction day report can also be used beyond the shell, even when the crane cameras are no longer in operation - but without the automatic process recognition.

Pre-filled input field of the construction day report

What the whole thing looks like in the end?

If you want to see how an AI generated construction report looks like in the end as an export, just click on this link!

Appendix: A small explanation of artificial intelligence

The algorithms developed by oculai consist, among other things, of various AI models, i.e. artificial intelligence. These algorithms are trained with large image data sets, in which recurring patterns and optical features of certain objects or surfaces are generalized and learned. In image classification, which is commonly used, an algorithm is shown so many photos of a "cat," for example, that the algorithm detects patterns in the pixel arrangements between the individual photos and would then recognize that cat in new photos. Based on this principle, oculai's AI models also learn to recognize processes from camera data.

Author:
Constantin Kauffmann