A range of solutions
EagleEye, HawkEye, and MicroRaptor all act on information you already have, whether it arrives live from a sensor or is retrieved from a recording. Choosing the best fit depends on your platform constraints: hardware capacity, bandwidth, or power. We have solutions that can run on fabric, embedded processors, or as software.
- Sensor performanceMaximise signal to noiseEagleEye
- BandwidthNotable eventsHawkEye
- PowerLow SWaPMicroRaptor
Our algorithms
EagleEye
EnhancementTemporal and spatial processing on the individual pixels, for platforms that need the most out of their sensors
Full specifications- Acts on
- Raw or compressed sensor output
- Output
- Enhanced frames
- Method
- Temporal and spatial
HawkEye
Event outputExtracts the moving edges in a scene and emits address-event representation data, compatible with existing event-based cameras. A target a few pixels across can be picked out of scene which would otherwise hide it.
Full specifications- Acts on
- Raw frames, or EagleEye
- Output
- AER, frames or overlay
- Method
- Moving edges only
MicroRaptor
Low SWaP luma enhancementA simpler model, cheaper to implement and effective in difficult light, for platforms where the power budget sets the ceiling.
Full specifications- Acts on
- Luma
- Output
- Enhanced luma
- Method
- Simpler model
On fabric, embedded processors, or as software
AXI IP block
EagleEye and HawkEye packaged as AXI-compatible IP, ready to drop into your own FPGA design alongside whatever else is on the fabric. MicroRaptor runs on fabric too, implemented on request rather than as a block you can order.
- EagleEye
- <1 µs, <1.5 W, <15%
- HawkEye
- <100 mW added, <1%
- MicroRaptor
- <200 mW, <5%
- Targets
- Ultrascale+ Zynq, Artix 7
TRS (Transparent Raw Sensor)
Our FPGA appliance, running EagleEye between an HDMI input and an HDMI output, so enhancement drops into a video chain you already have without touching the camera or the display.
- Runs
- EagleEye
- In and out
- HDMI
- Changes needed
- None either side
Cuvos Proof of Concept Camera
An FPGA camera that behaves like any other camera, with EagleEye and HawkEye both running on board.
- Runs
- EagleEye, HawkEye
- Complete unit
- under 10 W
- Size
- 25 x 20 x 7.5 cm
Microprocessor build
MicroRaptor on a small processor rather than fabric, which is where a low power budget usually lands. A separate product from the software application, not the same code on a smaller machine.
- Runs
- MicroRaptor
- On
- 4-core ARM
- Capture and processing
- 2.5 W
- System total
- 6.5 W
Software, on CPU or GPU
All three algorithms as an application rather than fabric, for pipelines with no FPGA in them and for footage already recorded. A recording has a lower ceiling than sensor output, since what the encoder discarded is gone for good.
- Runs
- All three
- Throughput
- 1080p60 real time
- On
- A laptop GPU
Figures are estimates on the targets named. The complete camera is a research and development unit.
Let's discuss your use case
Contact us for a more detailed discussion of the various trade-offs of our products and how they might fit within your application.