Metaspectral

Metaspectral is a Canadian startupstart-up developing computer vision software that can process hyperspectral data in real time using artificial intelligence (AI). Headquartered in Vancouver, British Columbia, Metaspectral offers fusionFusion, a cloud-based platform deploying deep learning models to analyze multispectral and hyperspectral imagery. Hyperspectral data contains a large number of spectral channels to discover information identifying the chemical composition and other properties of materials, enhancing computer vision applications. Metaspectral's software is being deployed in a range of industries, including aerospace, defense, agriculture, and manufacturing.

Use cases for Metaspectral's technology include the following:

• Recycling—detecting different materials to improve automated recycling systems.
• Space—earthSpace—conducting earth observation data analysis
• Defense—detecting CBRN (Chemical, Biological, Radiological, and Nuclear) materials
• Environment—precisionEnvironment—performing precision agriculture, and environmental monitoring, and wildfire mapping

The founders of Metaspectral, Francis Doumet (CEO) and Migel Tissera (CTO), originally started PixelDrive in 2018. The company offered a cloud storage platform that compressed uploaded photos using advances in machine learning. This led to Compression.ai in 2019, opening up the backend of PixelDrive to developers through an API and managed service. The company was approached by defense and space companies looking for efficient compression solutions for multispectral and hyperspectral data. In 2020, the company became Metaspectral. Building on in-house compression research on spectral data and work by other organizations, such as the MPEG consortium and the Consultative Committee for Space Data Systems (CCSDS), the company has since worked on developing its hyperspectral data analysis platform.

In July 2022, it was announced that Metaspectral would be working with HySpeed Computing (remote sensing data access and analysis company) to deploy OPTICA (Onboard Programmable Technology for Image Classification and Analysis), an earth observation payload on the International Space Station (ISS). The payload enables real-time compression, streaming, and analysis of hyperspectral data from low earth orbit (LEO).

On November 16, 2022, Metaspectral announced a $4.7 million seed round from SOMA Capital, Acequia Capital, the Government of Canada, and multiple angel investors, including Jude Gomila and Alan Rutledge. On March 20, 2023, Metaspectral announced it was joining Venture Catalyst Space, an Australian commercial space accelerator and incubator program. Venture Catalyst Space is run by the University of South Australia’s Innovation & Collaboration Centre (ICC) and is funded by the South Australia Space Innovation Fund. Kanyini, the first satellite designed and constructed in South Australia, includes a hyperspectral imaging payload. On April 12, 2023, Metaspectral signed a letter of intent with SkyFi, a company providing satellite imagery from a network of over 70seventy satellites. The letter makes SkyFi earth observation data available to users of Metaspectral Fusion.

Metaspectral is a Canadian startup developing computer vision software that can process hyperspectral data in real time using AI. Headquartered in Vancouver, British Columbia, Metaspectral offers fusion, a cloud-based platform deploying deep learning models to analyze multispectral and hyperspectral imagery. Hyperspectral data contains a large number of spectral channels to discover information identifying the chemical composition and other properties of materials, enhancing computer vision applications. Metaspectral's software is being deployed in a range of industries including aerospace, defense, agriculture, and manufacturing.

Metaspectral's deep learning software offers spatial and spectral analysis in real-time, extracting sub-pixel level data from hyperspectral sensors. It uses novel data compression methods to analyze the significant amounts of data output from hyperspectral systems. Users can label data and train sub-pixel AI models to analyze data using Metaspectral. MLOps agents monitor AI models for future improvements. The models can also be run at the edge as well as the cloud. Metaspectral software is hardware agnostic, compatible with any hyperspectral or multispectral sensor.

Fusion supports training with various neural architectures, including ResNets, convolutional neural networks, and autoencoders. The company has developed custom loss functions to train deep neural networks while addressing instability, convergence, and spectral unmixing. Metaspectral states that its algorithms outperform classical approaches, such as Adaptive Cosine Estimation (ACE), by up to 100x in minimizing false positives. The platform uses synthetic data generation processes to train target detection neural networks, enabling training using just a single spectral signature.

Use cases for Metaspectral's technology include:

• Recycling—detecting different materials to improve automated recycling systems.
• Space—earth observation data analysis
• Defense—detecting CBRN (Chemical, Biological, Radiological and Nuclear) materials
• Environment—precision agriculture, environmental monitoring, and wildfire mapping

The founders of Metaspectral, Francis Doumet (CEO) and Migel Tissera (CTO), originally started PixelDrive in 2018. The company offered a cloud storage platform that compressed uploaded photos using advances in machine learning. This led to Compression.ai in 2019, opening up the backend of PixelDrive to developers through an API and managed service. The company was approached by defense and space companies looking for efficient compression solutions for multispectral and hyperspectral data. In 2020, the company became Metaspectral. Building on in-house compression research on spectral data and work by other organizations, such as the MPEG consortium and the Consultative Committee for Space Data Systems (CCSDS), the company has since worked on developing its hyperspectral data analysis platform.

In July 2022, it was announced that Metaspectral would be working with HySpeed Computing (remote sensing data access and analysis company) to deploy OPTICA (Onboard Programmable Technology for Image Classification and Analysis) an earth observation payload on the International Space Station (ISS). The payload enables real-time compression, streaming, and analysis of hyperspectral data from low earth orbit (LEO).

On November 16, 2022, Metaspectral announced a $4.7 million seed round from SOMA Capital, Acequia Capital, the Government of Canada, and multiple angel investors including Jude Gomila and Alan Rutledge. On March 20, 2023, Metaspectral announced it was joining Venture Catalyst Space, an Australian commercial space accelerator and incubator program. Venture Catalyst Space is run by the University of South Australia’s Innovation & Collaboration Centre (ICC) and funded by the South Australia Space Innovation Fund. Kanyini, the first satellite designed and constructed in South Australia includes a hyperspectral imaging payload. On April 12, 2023, Metaspectral signed a letter of intent with SkyFi, a company providing satellite imagery from a network of over 70 satellites. The letter makes SkyFi earth observation data available to users of Metaspectral Fusion.

On April 27, 2023, Metaspectral secured $419,000 from the CleanBC Plastics Action Fund to support the development of Metaspectral's technology. The fund is intended for projects that reduce plastic pollution. Metaspectral was previously awarded £300,000 from the initial launch of the CleanBC Plastics Action Fund in 2021. In autumn 2023, Metaspectral began the beta rollout of Metaspectral Fusion, the company's cloud-based platform for real-time analysis of hyperspectral imagery using deep learning models.