BrainSIM: Brain-like Semantic Video coMpression
Dates: 10/2021 – 03/2025
Funded by: 2nd Call for H.F.R.I. Research Projects to support Post-doctoral Researchers
PI for FORTH: Effrosyni Doutsi
Funding: € 82,080
Summary:The BrainSIM proposes some novel brain-inspired video compression techniques, which are energy efficient while
ensuring a reconstruction quality that satisfies the human visual perception. There are several challenges
concerning the complexity and the power consumption of the latest video compression standards which have not
yet taken into account by the signal processing community. We propose that these challenges can be addressed by
artificial intelligence in order to avoid the exhaustive comparisons between sequential frames. We aim to release a
semantic video compression algorithm based on Convolutional Neural Networks (CNNs) that detect multiple
moving objects in order to do semantic bit-allocation. Another goal is to mimic the visual system as an intelligent
mechanism that processes the visual stimulus. This can be claimed as it consumes low power, it deals with high
resolution dynamic signals and the dynamic way it transforms and encodes the visual stimulus is beyond the
current compression standards. There have been a lot of efforts to understand the structure and functions of the
visual system and especially how the huge visual information is propagated and compacted through the nerve cells
into a sequence of electrical impulses, called spikes, before it reaches the visual cortex. There are very interesting
neuroscience models which approximate the neural behaviour and they have been widely used for image
processing applications including compression. The BrainSIM project searches the latest and more efficient
neuroscience models to design a groundbreaking video compression architecture that transforms the input signal
into spikes. Last but not least, we aim to analyse the spikes using the Recurrence Quantification Analysis; for the
analysis of complex dynamic structures. The BrainSIM is expected to improve the performance of the latest stateof-
the-art and draw new horizons for image processing community.
Visit Website