Using Machine Learning for Geoglyph Detection

PNAS has an interesting paper on using AI and Machine Learning to try and identify new Nazca Pampa geoglyphs in the Peruvian Nazca Desert. It’s a fun little lunchtime skim with a few nice images of some of the newly found geoglyphs.

PNAS: AI-accelerated Nazca survey nearly doubles the number of known figurative geoglyphs and sheds light on their purpose

Converting Video to Image Sequences with ffmpeg

Another little aide-memoire masquerading as a blog post; the specific incantation required to turn a video into a named sequence of images with ffmpeg is…

ffmpeg -i <path to source video>  -vf fps=<frames per second> <output directory/basename->%d.png

…where…

<path to source video> is the path to the video file in question.
<frames per second> is the number of frames per second of footage to extract. This can be less than 1 if you wish to extract at a lower rate than one frame per second.
<output directory/basename->%d.png is a composite instruction to create files in the directory ‘output directory’, for these files to have the prefix ‘basename-‘, and for these files to have an incrementing count appended to the end. It also specifies that the output files should be in the ‘.png’ format.