david-durst / magmacam Goto Github PK

This repo contains applications and demos for embedded FPGA systems developed using the magma HDL.

Here is a breakdown of the various directories.

This directory contains Magma programs written for the Lattice ICE40 IceStick architecture.

The main program which defines a top level module (main). It uses the ArduCAM circuit (defined in arducam.py), which exposes the SPI interface of an ArduCAM Mini 2MP peripheral for initiating a capture and receiving image data, and the Process circuit (defined in process.py), which processes RGB pixel data converting it to grayscale values.

BNN digit classifier adapted from https://github.com/MIT-HAN-LAB/BNN_IceStick. Instead of the loading the image into ROM at initialization time, the circuit has a RAM for storing image bits so that the input of the BNN may be wired to another circuit (e.g. the output of the ArduCAM).

TODO: It would be good to document how to run these tests (using the logic analyzer?) Also, maybe these should be moved to a tests directory. We could also consider writing some tests for these using the Python or CoreIR simulator, so we could run them on Travis.

Test program for the STEN aetherling module (see below).

Test program to downscale a 320x240 ArduCAM image to a 16x16 using the Downscale aetherling module. Currently too large to run on the IceStick.

Test program which feeds an image stored in ROM to the RAM in the pipeline.py circuit and runs the BNN classifier.

This folder contains high level magma circuits defined using the aetherling library. It contains the STEN module (in test_linebuffer.py) which downscales a 4x4 image to a 2x2 using a window size of 2x2 and summing over the pixel values in each window to obtain each new output pixel value, (TODO: Should this average the pixels? Are they grayscale?) and the Downscale module (indownscale.py) which is designed to downscale a 320x240 image (taken by the ArduCAM) to a 16x16 (to feed into the BNN classifier) using a window size of 20x15.

This folder contains some miscellaneous python scripts. bmp2png.py creates a png image from a csv file of 16-bit RGB pixel values (as bytes), bytes2jpeg.py creates a jpg image from a a csv of raw JPEG bytes, and gray2img.py displays a grayscale image from a csv file of grayscale values. monitor.py is a simple program to receive UART data from the IceStick. (TODO: Would be good to note what these are used for, such as example use case. E.g. the bmp2png can be used with the logic analyzer)