HEA BLO INS AMLD Conference Video

Video: Pix4D presents at an Applied Machine Learning Day

Pix4D presented at an Applied Machine Learning Day to share the ideas, techniques, and tech behind drone cell tower inspections

The Applied Machine Learning Days (AMLD) is an event hosted by the world-renowned technical university EPFL in Lausanne. AMLD features keynote speakers, workshops, and networking opportunities for professionals and students in the field. 11 years ago, Pix4D emerged as a start-up from the EPFL, and it is the continued good relationship between them that saw Pix4D speak about its work with Machine Learning and Artificial Intelligence (AI).

Pix4D’s presentation at this event was spearheaded by Daniel Lopez and Andrea Dotti from the Pix4D research and development teams. They shared a case study from two of Pix4D’s products, the PIX4Dscan flight app and the online inspection platform PIX4Dinspect, that uses AI to automate the process of inspecting assets in different industries, including in telecom. These products use drones and drone imagery to create 3D models of objects with integrated AI that automates the inspection process to save time and provide analytical data. In some cases, they have made tower inspections 60% faster. Learn more by watching Andrea and Daniel’s presentation at AMLD:

The products behind drone telecom inspections

PIX4Dscan is a drone flight planning application that has custom-designed missions for cell tower inspections. As a result, it has flight plans that are designed to be adaptable according to the type of asset being inspected - whether it is in an urban environment, a camouflaged tower (eg hidden in a fake tree), or free-standing in a rural field. The flight plans can be adjusted for nearby obstacles, such as trees or fencing. Once the pilot is ready, the drone will fly an automated path, capturing the minimum number of images needed to create a 3D model. Once complete, the results are uploaded to PIX4Dinspect and automatically processed.

IMA INS AMLD Conference Video 2
Pix4D has a specialized set of products for Telecom inspections that uses AI

PIX4Dinspect is a cloud-based inspection platform. When images are uploaded to PIX4Dinspect for processing, the software uses 14 specific automation and AI algorithms to automatically recognize objects and measure them. For example, it can independently identify antennas on a cell tower and measure their height, length, azimuth, and downtilt. These details are then automatically compiled in a report. This saves a huge amount of time for inspectors and engineers, who can look at the processed results and see which maintenance requirements have already been identified. PIX4Dinspect is cloud-based, which allows for multiple projects to be processed simultaneously, easy sharing between stakeholders, and the ability to host a network of inspections (even from multiple countries) all in the same place.

The future of artificial intelligence in drone inspections

Together, these products save users significant time and energy. It is also highly accurate, where the precise measurements add to the value of using aerial inspections with drones. No longer is it necessary for an engineer to climb a tower for inspections. In addition, the entire workflow can take just an hour, as Daniel explained in Pix4D’s talk. They discussed a use case from SkyVue in Cameroon, where the entire inspection workflow was reduced from taking at least 24 hours to just 1 hour, for both data collection and processing. The value of this is growing, with telcos worldwide taking an interest in this, especially because it is a scalable telecom inspection workflow that can be replicated at multiple sites in a single day.

The growing interest in drone inspections for the telecom industry is driving Pix4D’s investment in this area. Already, PIX4Dinspect is available in English, Japanese, French, Spanish, Portuguese, German and Italian. Free trials are available for people to test how they find the workflow and see the analytics for themselves (demo datasets are available). With the ever-increasing capabilities of drones and AI, this industry will continue to grow and aerial inspections with UAVs could soon become the norm.


Enhance inspections with artificial intelligence
Specialized solution for your workflows

Related articles