This solution connects the front camera and side view cameras to TDA4-A, and transfers intermediate results to TDA4-B for further processing. The multiple result fusion is then implemented in TDA4-B and the final result is outputted, as shown in Figure 3-2.

The difference between this scenario and the previous scenario is the result of TDA4-A being transferred through the PCIe TX node. Deep learning inference intermediate results are received by the PCIe RX node in the TDA4-B node, and those results are used as the source data for further processing in TDA4-B. The surround view camera results are input to the fusion node along with the deep learning result. Final results are outputted from TDA4-B.