Support Center
The IMX501 CMOS rolling shutter sensor features 12.3 MP resolution at 4056 x 3040 px and on-sensor AI processing, enabled by its integrated ISP (Image Signal Processor), DSP (Digital Signal Processor), and on-chip SRAM. This on-sensor AI processing eliminates the need for expensive PC host resources, reduces latency, and lowers power consumption. Additionally, since AI processing and computation occur directly on the sensor and camera, no online cloud connection is needed. These AI models run entirely offline, locally on the camera.
Object detection
Sony's IMX501 Intelligent Vision Sensor
Triton Smart Camera: Beyond Small and Light
IP67 Protection
-20°C to 55°C Ambient
Active Sensor Alignment
Power over Ethernet
The Triton Smart Machine Vision camera, weighing just 68 grams and measuring 29 x 29 mm, is designed for reliable operation in challenging industrial environments. Its electronics and AI sensor are housed in a durable, sealed two-piece aluminum case secured by four M2 screws. The camera features robust M12 and M8 connectors for Ethernet and GPIO, resistant to shock and vibration. An optional IP67 lens tube offers dustproof and water-resistant protection without an external enclosure. The Triton operates efficiently within a wide temperature range from -20°C to 55°C ambient.
Triton Smart Camera Model with Sony IMX501
| Models | MP | Resolution | FPS | Sensor | Format | Pixel Size | Shutter | Lens Mount | Chroma | Interface |
|---|---|---|---|---|---|---|---|---|---|---|
| TRS123S-CC | 12.3 MP | 4056 x 3040 px | 7.9 fps | Sony IMX501 CMOS | 1/2.3" | 1.55 µm | Rolling | C | Color | M12 GigE Vision |
Triton Smart AI Video Benefits
Learn how the Triton Smart delivers fully offline edge AI, supports a free open-source AI development toolkit with no licensing or subscription fees, and combines a compact design with IP67 protection, PoE, secure M12/M8 connections, and a competitive price of $495 USD. We also explain its dual-ISP architecture and how developers can build their own AI models.
Triton Smart Camera: Processing Pipeline
The Triton Smart features two ISPs (Image Signal Processing): one located on-sensor for the AI engine and the other on the camera’s FPGA for regular image processing.
The IMX501 sensor captures high dynamic range, low noise RGB images, which are then sent to both the sensor’s ISP and the camera’s FPGA ISP for separate processing. The AI ISP pre-processes the image for the AI engine into an input tensor, which then outputs the inference results in the output tensor. Simultaneously, the image is sent to the FPGA ISP, where it is processed as it would be in a standard machine vision camera. The input and output tensors are embedded as chunk data and combined with the image from the FPGA ISP before being sent to the host PC.
To simplify AI model development for the Triton Smart, LUCID provides a Google Cloud VM preconfigured with an open-source, PyTorch-based training framework, Docker, and the Triton Smart Model Converter for model training, quantization, conversion, and deployment to the camera.
Building and Viewing AI Models for Triton Smart
Open Source AI Development Toolkit
Open Source AI Development Toolkit – LUCID provides access to a VM image via Google Cloud Platform (GCP) to simplify AI model production for the Triton Smart. This VM contains all necessary AI training tools and is preconfigured to create Triton Smart compatible models. This workflow provides greater flexibility and control over your AI models without any licensing or subscription fees.
ArenaView MP
Once an AI model is created using the AI development toolkit, LUCID’s free ArenaView MP software is used to upload the model to the IMX501 sensor on the Triton Smart camera. Once the camera is running, users can use pre-compiled code to view real-time images with AI information overlays. This software operates entirely offline and locally, for added security and privacy. Users can also use the Arena SDK to process the results of the AI engine.
Triton Smart Camera Model with Sony IMX501
| Models | MP | Resolution | FPS | Sensor | Format | Pixel Size | Shutter | Lens Mount | Chroma | Interface |
|---|---|---|---|---|---|---|---|---|---|---|
| TRS123S-CC | 12.3 MP | 4056 x 3040 px | 7.9 fps | Sony IMX501 CMOS | 1/2.3" | 1.55 µm | Rolling | C | Color | M12 GigE Vision |
