SoM and carrier board
The qSmartAI80_CUQ610 vision kit is an Edge AI vision kit based on the Qualcomm QCS610 processor with a 4K ultra low light camera based on the Sony STARVIS IMX415 sensor. This is a complete ready-to-deploy vision kit using which product developers can right away develop their AI-based embedded vision device.
To learn more about the details of the kit components, the AI stacks that come with it, and its key differentiators, please visit the article Qualcomm QCS610 AI vision kit with e‑con's low light camera module.
Yes. e‑con Systems will be the design partner for carrier board customization. e‑con would provide the final design files (Gerber and BOM) for the customized carrier board. You have the liberty to choose your manufacturing partner. However, e‑con can also help you with the same.
Yes. Being a complete solution provider, e‑con Systems shall be able to design the custom box with Qualcomm processor including the carrier board, camera, and enclosure according to your specifications.
The duration required to develop a custom AI vision box depends on the complexity of your end application. Typically, it ranges from 10 weeks to 20 weeks. For a detailed timeline for building a custom AI vision kit using QCS610, please write to us at camerasolutions@e‑consystems.com.
VVDN Technologies manufactures the SoM that comes embedded in qSmartAI80_CUQ610.
Yes. e‑con Systems shall supply the SoM along with the carrier board and camera to its customers.
Yes, of course. e‑con's portfolio of cameras includes IP67-rated cameras that are tested for quality and endurance. Apart from that, e‑con has helped several customers build custom IP67-rated enclosures for its camera modules. Given below are two of e‑con's off-the-shelf products that come with an IP67/66 rated enclosure (as mentioned before, e‑con can design an IP67 rated enclosure for any of the products in its portfolio. Please visit the Camera Selector to have a look at our complete list of cameras by features):
Being a Qualcomm partner, e‑con Systems shall help design, develop, and manufacture the carrier board for customers according to their specifications. If e‑con is customizing the carrier board for you, we shall share the schematics for the custom design.
This AI vision kit uses a 4 lane MIPI-CSI2 interface to connect the camera and the processor. To learn more about MIPI cameras, please have a look at the article What is a MIPI camera? How does MIPI camera work?
This vision kit has a Sony STARVIS IMX415 ultra low light camera module. The Sony STARVIS series of sensors is known for its high sensitivity and superior low light performance. From Exmor to Exmor R to Exmor RS and STARVIS, Sony sensors have gotten better over time. To learn the differences between various types of Sony sensors, please read the article Sony Exmor vs STARVIS sensors - a detailed comparison.
As mentioned before, the camera is using the MIPI interface to connect to the Qualcomm host. And the connection is done using a high quality coaxial cable.
This vision kit comes with a 30 cm coaxial cable by default. However, e‑con supports cables up to a length of 1 meter, which requires a customization. For any help in altering the cable length of this vision kit, please write to us at camerasolutions@e‑consystems.com. You could also visit the OEM camera customization page to learn more about all the customization services offered by e‑con Systems.
Currently the qSmartAI80_CUQ610 kit hardware supports up to 3 cameras. Software support is provided to connect and integrate 2 cameras for now. However, driver support for the 3rd camera can be taken up as a customization request.
Yes. This kit supports 1 x USB3.1 and 2 x USB 2.0 interfaces. Hence you have the choice to connect any USB camera to this kit. Please visit the USB3 cameras and USB2 cameras pages to have a look at e‑con Systems' complete portfolio of USB cameras.
Currently, this kit comes with driver support for MIPI-CSI2 cameras. Supporting GMSL2 cameras is very much possible. In case you are looking to customize the camera to connect to the host using the GMSL2 interface, please write to us at camerasolutions@e‑consystems.com with your detailed requirements. Meanwhile, please have a look at our complete portfolio of GMSL cameras here.
The camera comes with an M12 lens holder. Hence, you shall be able to use any standard S-mount lens (M12 x 0.5) with the default lens holder. Please write to camerasolutions@e‑consystems.com for lens suggestions.
It is possible to support a C-mount lens holder as a hardware customization effort. Please visit the OEM camera customization page to have a look at all the customization services we provide. You could also write to us at camerasolutions@e‑consystems.com for any help with camera customization.
The default lens supplied with the camera has an IR cut filter that filters light in the InfraRed spectrum. If you wish to use the camera in NIR conditions, we recommend you to test it using a different lens that passes IR through it. For any help in choosing a different lens, kindly write to us at camerasolutions@e‑consystems.com. To learn what an NIR camera is and the various NIR cameras offered by e‑con Systems, please do have a look at the article What is NIR imaging and how do NIR cameras work?
Yes. This camera uses the internal ISP of Qualcomm QCS610.
Yes. e‑con would support any finetuning required on the Qualcomm ISP to meet your specifications. ISP tuning is one of e‑con's key areas of expertise. Our ISP tuning capabilities have helped us win numerous deals against some of our top competitors.
Yes. e‑con Systems shall be able to support the integration of any image sensor with the Qualcomm AI vision kit. However, e‑con is already having a no of cameras with a wide variety of sensors from Sony, Onsemi, and Omnivsion. Please visit the Camera Selector to have a look at e‑con's complete portfolio of cameras based on these sensors.
Yes. This would be a custom solution. e‑con Systems shall be able to design a custom camera with the necessary drivers.
Yes. It is possible to connect cameras with different sensors to a single board.
Camera synchronization shall be provided as a customization service. e‑con Systems offers various camera solutions with multiple synchronized cameras. The multi-camera solution with 180 degree stitching for NVIDIA Jetson AGX Xavier is an example for this.
Yes. This vision kit comes with QCS410 based SoM as well. We will be launching the same with required software support soon. Please write to camerasolutions@e‑consystems.com in case you are looking for a QCS410 based solution.
Yes. We have a camera solution available for DragonBoard 820c development board which is based on the Qualcomm Snapdragon 820E embedded platform with the 64-bit Qualcomm Kryo quad-core CPU. Here is the link to the product: e-CAM51_CU96 - 5MP camera for Qualcomm DragonBoard 820C.
Every application is unique and the host processor has to be chosen by taking into consideration a multitude of factors. It is recommended to have a discussion with our camera solution experts who would be able to help you with this. Please write to camerasolutions@e‑consystems.com to set up a consultation call with our experts.
NVIDIA offers a wide variety of processors including NVIDIA Jetson Nano, Xavier NX, TX2 NX, AGX Xavier, AGX Orin, etc. Each of them comes with a wide range of performance starting from 472 GFLOPS to all the way till 275 TOPS. To understand which of these platforms you are using, we recommend you to setup a discussion with our camera solution experts who would be able to give you the necessary details of each processing platform. They can also help you with choosing the right platform for your embedded device. For any such help, please write to camerasolutions@e‑consystems.com.
All the 3 series of processors come with a varied range of performance. i.MX 8M Plus - with its Neural Processing Unit operating at up to 2.3 TOPS - is suitable for low end AI applications such as face detection and recognition. Please visit the NXP cameras page to have a look at e‑con's portfolio of NXP cameras including those compatible with i.MX 8M Plus.
As mentioned before, NVIDIA Jetson processors are suitable for high end AI applications that require as many as 40+ cameras. For instance, Jetson AGX Orin - which is the latest addition to NVIDIA's portfolio of processors - comes with the highest performance of 275 TOPS (with 64 GB memory). Hence, NVIDIA processors are typically the best for high end AI applications that have a heavy computational load.
On the other hand, the latest Qualcomm processors are suitable for low end/medium end AI applications such as face detection, face recognition, object tracking, and people counting.
To know how to choose the best processing platform for your AI application, please write to us at camerasolutions@e‑consystems.com.
Yes. Qualcomm QCS610 supports H.264 encoding up to a resolution of 4K@30fps.
Software / Application
Yes. e‑con Systems provides the complete ecosystem for customers to develop and deploy their own algorithms on top of the kit. We shall provide the necessary integration support for the AI SDKs that come with the kit. Please refer question #38 to learn in detail about the AI SDKs available in qSmartAI80_CUQ610.
e‑con Systems shall be able to help you with any necessary customization on the BSP side.
Gstreamer based sample application is provided to test and evaluate this smart camera kit.
Yes. Please refer question no #38 for more info.
Yes. Both Caffe and Caffe2 frameworks are supported. Please refer question #38 for more info.
Following are the AI stacks supported by qSmartAI80_CUQ610
- Qualcomm Neural Processing SDK for AI: This Qualcomm AI-based Neural Processing SDK enables product developers to run one or more machine learning and neural network models trained in Caffe/Caffe2, ONNX, or TensorFlow.
- Neural Network API (TFLite models); The Neural Network API provides acceleration for TensorFlow Lite models with supported hardware accelerators including a GPU (Graphics Processing Unit), DSP (Digital Signal Processor), and NPU (Neural Processing Unit).
- GStreamer plugin (DLC models): The GStreamer DLC plugin (qtimlesnpe plugin) helps to smoothly integrate a .dlc to the available GStreamer plugins.
- GStreamer TFLite plugin: This GStreamer plugin (qtimletflite plugin) exposes TensorFlow Lite capabilities to GStreamer. While the pre-processing functionality supports downscale, color convert, mean subtraction, and padding, post-processing supports models such as classification, detection, and segmentation.
Please refer Machine Learning on QCS610 - Qualcomm Developer Network for more details.
Yes. Qualcomm provides support for Microsoft Azure machine learning and other Azure services.
Android support will be available once Qualcomm provides it.
Yes. e‑con is ready to provide this kit on a returnable basis for evaluation given that the customer is having the right use case and mass volume requirements. Please contact sales@e‑consystems.com if you are looking to avail this offer. However, it is to be noted that providing the kit is completely e‑con's discretion.
e‑con is willing to provide this kit on a returnable basis for evaluation given that the customer is having the right use case and mass volume requirements. Please write to sales@e‑consystems.com for more details on this. However, e‑con holds the right to take a call on whether a prospect gets a free sample or not.
For any pricing related information, please contact us at sales@e‑consystems.com.
Yes. 1000 is the MOQ for mass volume production.
As the case with all our camera modules, e‑con Systems shall supply this Qualcomm kit for a minimum of 5 years.
As mentioned in the previous response, e‑con Systems shall supply the vision kit for a minimum of 5 years.
Qualcomm QCS610 is suitable for any mid-tier AI application. Examples for such applications include telepresence robots, service robots, smart carts, digital signages, smart traffic systems, etc. If you wish to have a brainstorming around how to build your next camera-based embedded device using Qualcomm QCS610 and e‑con cameras, please feel free to write to us at camerasolutions@e‑consystems.com.
All off-the-shelf products from e‑con Systems come with a 3-year warranty which applies to qSmartAI80_CUQ610 as well. Please have a look at our warranty page for more details.