SpeckCollect

A simple module that collects events from a SynSense Speck2fDevKit and creates temporal frame representations of the collected events. Users can set the time window to collect events over, essentially the 'framerate'. Simply counts the number of events incurred for each pixel (128x128).

Currently, only considers positive events - functionality to be improved in later iterations to include both positive, negative, only negative, and merged events (v0.2.0).

Usage

Usage is very simple. First, install all the dependencies. Note, that currently samna required python <=3.11 so it's recommended you create a new conda environment to avoid issues with other libraries and dependencies.

# Recommended: create new conda environment
conda create -n speckcollect
conda activate speckcollect

# Install dependencies
pip install speckcollect

Download the repository and navigate to the repository directory.

# Clone repository
git clone git@github.com:AdamDHines/SpeckCollect.git

# Set current working directory
cd ~/SpeckCollect

To run the SpeckCollect module, simply use the main.py script which consists of a few arguments.

• --time_int - sets the time in which to collect events over (default 0.033s or 30fps)
• --directory - set the directory to save the data to (default ./speckcollect/data)
• --exp - set a name for the experiment (default exp)

# Run the SpeckCollect over a 1s timebin and set the experiment name to TEST001 
python main.py --time_int 1 --exp TEST001

The output will include a TEST001.npy file of the raw events collected and a folder TEST001 which contains .png files of frames for the temporal representation of events.