Porting Yolov7 Tiny Pytorch AI model onto Ability’s edge AI camera
#MalgnST, Ability’s valued partner in Korea, was very impressed with the short turnaround time it took them to successfully porting their #Yolov7-Tiny (Pytorch) AI model onto Ability’s edge AI camera. The surveillance camera was installed on the side of a street looking at a cross junction that is about 80 meters down the road. In the PoC experiment the team at MalgnST surprisingly found that their AI model running on the edge camera had no problem to detect and recognize different objects of pedestrian, bicycle, and vehicles even in the dark evenings and at a remote distance of 80 meter. The image captured and used for the AI inference is 960 x 960 pixels and the AI performance is about 20 frames per second.