Hi,
I'm experimenting with the semantic segmentation use case on the TDA2X EVM board. As a test, I want to draw bouding boxes around detected objects (etc. cars).
I've placed the BB drawing APIs in Alg_tidlpostproc_cpy after the dma copy.
I'm using Draw2D_drawRect to draw BBs on the output. I also do Cache_wb on the output buffers after the draw API.
What I'm observing is some distortion around the Bouding box region in the image. Is there something else I need to take care of?
Attaching a video which shows the effect.