DPM is a learning-based object detection FPGA IP core, developed for embedded vision applications. It uses a star-structured part-based model, defined by a root filter plus a set of parts filters and associated deformation models. This approach leads to a flexible and efficient object detector that achieves state of the art results on the PASCAL and INRIA person datasets.
User-defined classifiers can be loaded via software API.
In order to detect objects in an arbitrary range, the core accepts in input a pyramid of images.
Detection results are stored into an internal buffer, ready for the host controller read-out.
The core is provided with drivers and an embedded software library in C code implementing post-processing functions necessary to group together overlapping detections.
The DPM Demo Kit allows user to quickly evaluate and experiment with DPM core on the Xilinx Zynq-7000 SoC ZC706 Development Board with XC7Z045 FFG900-2. The DPM demo kit relies on a companion DVI I/O FMC daughter card (AES-FMC-DVI-G) from Avnet Electronics Marketing; support for HDMI I/O FMC Module (AES-FMC-IMAGEON-G) is available on request.
This free and pre-verified design includes evaluation DPM core and hardware design files prepared for Xilinx Design Suite 14.7 (including Xilinx ISE, Xilinx Platform Studio - XPS, System Generator for DSP and the SDK). Demo Kit deliverables include the complete software DPM demo application, which can be customized and upgraded to closely fit to specific requirements. Demo kit is available for Vivado® Design Suite on request.
See the Demo Kit in action in the video below: