Deep Learning & OBIA for Car detection
This video shows the results of an eCognition Developer analysis of an aerial image to detect cars. We used a combination of Deep Learning functionalities within eCognition Developer and OBIA (Object-based Image analysis) to detect cars only having optical data (RGB). First we created / trained and applied our own Deep Learning CNN model within eCognition. The resulting heatmap was then used to create image objects and delineate and locate cars using an OBIA.
Input:
High spatial resolution optical data (RGB)
Output:
Car outline
Car location as point
We have a FREELY available eCognition Deep Learning course: https://learn.trimble.com/learn/cours...
You can use the FREE trial version of eCognition to play around:
https://www.geospatial.trimble.com/ec...
If you have more questions, ask them in our FREE Community:
https://support.ecognition.com/hc/en-...