Hello there, we are from Digitronix Nepal (FPGA based design company). We have worked with Zynq families from past few years, we do have skillset on VHDL/Verilog, High Level Synthesis ( C/C++ OpenCV),Embedded C (for SDK programming for software application) along with Tcl and Bash.
We previously worked for "Real Time Canny Edge Detection with HLS and Zynq FPGA", this design was inspired from xapp1167 and targeted ZC702 and ZedBoard.
Previously we have worked on AES Encryption IP Design and Verification for Zynq FPGA.
Aside of this we do have experience on Image/Video Processing with Pynq FPGA (Python+Zynq) FPGA, on talking about this; PYNQ enable fast prototyping of Computer Vision Algorithm in FPGA's. So you might need to review this platform to cause most of backend is already ready, what need to do is just write CNN on Python with Jupyter Interface.
If you like to know more about is, we have 10 Courses on Udemy with ~1000 Students on VHDL/Verilog Programming, Zynq Development with VIVADO, Zynq Ultrascale+MPSoC Development, FPGA Design with HLS, FPGA Design with SDSoC etc.
Actually, Your Budget is to low, so please revise it. We just have purpose our milestone with respect to your budget, which will need to be revised.
thanks!