WildCamera

---

Planning phase

<img align="right" width="400" src="Images/Will 20240827_142415 Rotation Corrected.jpeg" alt="Will Robertson"/>

1. Requirements Analysis

• Why:

  • Ecologists and zoologists identify shortcomings with commercial wildlife camera systems:
    • Limited reliability particularly with small, fast moving mammals
    • Camera alignment is often time-consuming and difficult
    • Extensive driving to check on status of cameras
    • Too slow for fast moving animals
    • Limited ability to focus nearby resulting in poor image quality for small animals
    • Limited optics can only be changed by sticking extra lenses onto the existing optics with resulting decline in image quality or by time-consuming dismantling, removal of glue, adjustment and rebuilding
    • Batteries are often primary cells - rechargeable batteries need changed for charging frequently (e.g. weekly)
    • Lack of flexibility adding new features is extremely difficult or impossible, users have no access to electronic or software design
    • Reproducibility because of differences in reliability is is difficult to compare studies carried out with different makes and models of cameras
    • Waterproofing is often inadequate for reliable long term use in damp environments
    • High cost is a major barrier to research but manufacturers are driven by maximising profits
    • Expensive proprietary tie-in for example to specific mobile networks - increasing long-term costs
    • PIR sensors work for large mammals but reliability is an issue for small mammals, reptiles or insects
    • Availability preferred makes and models may not be available in the future
    • Dated technology large amounts of proprietary firmware and electronics and limited competition make commercial wildlife camera manufacturers slow to adopt new technologies and unable to catch up with rapidly advancing and highly competitive image sensor and processor technology
    • Inability to differentiate between animals and branches moving in the wind resulting in storage of large volumes of irrelevant video
    • Limited file system typically limited to 64 GB which fills up rapidly with uncompressed video
    • Technical data is often sales-focused, omitting important information or known limitations
    • Image quality varies between models and day/night

• Objectives:

  • Video from animals nearby and at a distance via a flexible choice of image sensors and optics
  • Activation via a flexible choice of activation sensors (PIRs, always-on camera, etc.)
  • Data Capture from a range of optional add-on sensors
  • Power Management via an active mode and a low power sleep mode
  • Power Supply from a flexible choice of solar panels, battery chemistries and other supplies
  • Robustness through high quality watertight enclosures
  • Re-Use of 2nd hand photovoltaic modules (solar panels) and rechargable batteries
  • High Quality and Exceptional Value through drive for excellence amongst image sensor manufacturers
  • Exceptional Capabilities, Flexibility and Value through drive for excellence amongst processor manufacturers serving fast-growing industrial automation markets
  • Independence in the Field via data download within WiFi range, solar power and optional LTE (mobile network) and sub-GHz
  • Open Source users can freely build and use the system and add open source functionality to make available to others
  • Hardware Independence to allow rapid adoption of leading-edge hardware from leading manufacturers - acheiveing optimum value for money and eliminating risk of proprietary tie-in
  • Reproducible researchers can share and precisely reproduce each other's setup
  • Future Proof designed to be widely available in the long term
  • Agile Designed to take advantage of future advances in hardware and software
  • Optional Add-On Modules to add additional functionality
  • Separate Modules in separate boxes connected by flexible cables or in one box
    • Camera small with complete freedom of positioning
    • PIR multiple PIR (Passive Infra Red) sensors can be mounted some distance away if needed - wide choice of high sensitivity, wide angle, etc.
    • AIR multiple AIR (Active Infra Red) sensors if needed in addition to or instead of PIR where needed
    • Processor larger than camera, may have optional hardware modules added
    • Battery likely the largest module - may be some distance from processor and camera
    • Solar Panel(s) may be mounted some distance from battery and processor to gather light
    • IR LEDs can be positioned independently of camera, light output can be set in software, IR wavelength can be chosen
    • Visible or IR Flash where required

---

<img align="right" width="400" src="Images/Goedele - September 2024 - Cropped - P8280995.jpeg" alt="Goedele verbeylan" />

2. Key Technologies

• Camera and Optics :
  • Focal Length 20mm to infinity
  • Focusing
    • Manual lens adjustment
    • Fixed via user interface
    • Autofocus via a choice of algorithms
  • Daytime Combined Red, Green, Blue and NIR (Near Infra Red)
  • Nighttime High sensitivity monochrome and IR (Infra Red)
• Processor Options
  • High Quality, High Flexibility
    • Image Quality High
    • Choice of Processor Wide
    • Choice of Camera Modules Wide
    • Solar Panel and Battery Larger
    • Firmware Full flexibility of Linux operating system, v4l2, libcarmera, Python, Rust, .Net, C++, etc.
    • Dependency Minimal dependencies between software and hardware
  • Low Power
    • Image Quality Limited but sufficient
    • Choice of Processor More limited
    • Choice of Camera Modules More limited
    • Solar Panel and Battery Smaller
    • Firmware ThreadX, FreeRTOS, Zephyr, bare metal, C, C++, MicroPython
    • Dependency More dependencies between software and hardware
• AI
  • Built-in NPUs (Neural Processing Units) allow neural networks to be efficiently implemented on the camera
• Activation Sensor
  • PIR Choice of
    • General Purpose
    • Narrow Angle
    • Wide Angle
    • High Sensitivity
    • Adjustable Sensitivity

---

3. Hardware Selection

• Candidate Processors
  • High Quality, High Flexibility
    • NXP i.MX 8M Plus Dual ISPs: up to 12MP and 375MPixels/s
    • NXP i.MX95 Pre-production: 500 Mpixel/s MIPI-CSI and ISP (2x 4-lane, 2.5 Gbps/lane)
    • ST STM32MP2 Series On ST roadmap. CSI-2 #1 (5 Mpixels @30 fps with ISP), #2 (1 Mpixels @15 fps no ISP)
    • Raspberry Pi 4 (RPi 4) SBCs High power consumption and lack of low power sleep mode are a significant problem.
    • Raspberry Pi 5 (RPi 5) SBCs Lack of hardware video encoder, high power consumption and lack of low power sleep mode are a very significant problem.
    • Raspberry Pi Zero 2 (RPi Zero 2) Power consumption, lack of sleep mode and speed of video encoder (19 FPS for 1080p) are a problem but may be worth testing.
    • Raspberry Pi Zero (RPi Zero) Video encoder is too slow (1.2 FPS 1080p).
    • Chips Integrating Mali-C55 ISP
      • Renesas RZ/V2H Mali-C55 ISPC
    • Chips Integrating Mali C52 ISP
      • Possibly in the longer therm
    • Chips Integrating Amlogic C3 ISP
      • Currently in development by Amlogic
  • Low Power
    • ST STM32N6 Series μCs ISP, MIPI CSI-2, H264 hardware video encoder, Neural-ART NPU
    • NXP S32V2 Embedded ISP for HDR, color conversion, tone mapping, etc. enabled by ISP graph tool

<video controls src="https://github.com/user-attachments/assets/248f8eea-008a-45c1-bc7c-6e400344eacd" alt="Eliomys quercinus © Goedele Verbeylen"></video>

• Candidate Cameras and Image Sensors
  • Sony PRi Camera Module 3 IMX708, Soho Enterprise Camera Modules or RPi High Quality Camera IMX477
  • ST BrightSense RGB and NIR e.g. VD1940 5.1 MPixel, VB1740 2.7 MPixel
    • VD56G3 monochrome, VD66GY RGB and VD16GZ RGB-IR - 1.5 MPixel - in production
    • VD5943 monochrome and VD1943 RGB-IR 5 MPixel in pre-production
  • ST BrightSense NIR Optimised e.g. VG5761 2.3Mp HDR ultra low-noise
  • OmniVision OV64A40 e.g. VG5761 2.3Mp HDR ultra low-noise
  • OmniVision OX05B1S RGB-IR 5Mp image sensor
  • Global Shutter undistorted photography of fast moving events - Sony and ST
  • Others Other image sensors supporting libcamera
• Candidate Enclosures
  • Serpac RB Series IP67, UV stabe
  • Scame Parre S.p.a SCABOX Series IP56
• Activation Sensors Passive Infra Red (PIR) or other technologies for specialist applications.
• Storage MicroSD card or internal flash memory (eMMC).
• Data Communication Wi-Fi (preferred) for video, photo and data download to a laptop within range (Bluetooth may be too slow for video).
• Wakeup Communication Low power protocol e.g. Matter IoT, Bluetooth LE, Zigbee or sub-1GHz IEEE 802.15.4 to send wake-up signal to waken up WiFi above.
• Power Supply solar panel(s), external power source.
• Battery Li-Ion for moderate temperatures, LiFePO4 provides power down to -20 °C and can be charged down to 0 °C, PbA for lower temperatures.

---

4. Optional Add-On Modules

• RFID / PIT to read ID from tagged animals
• LASER Spot Alignment to make the system quicker and easier to align in the trees
• 4G LTE Remote monitoring and control via mobile phone network via LTE (Long-Term Evolution) - successor to GSM to communicate over mobile phone network
• Thermal Imaging Camera / FLIR image sensors extending further into the IR (Infra Red)
• Temperature Sensor small and robust - multiple temperature sensors can be connected and can be some distance from the processor
• Humidity Sensor
• Ambient Light Sensor
• Microphone
• Ultrasonic Microphone
• Heterodyne Detector To shift ultrasonic bat calls t