Video Album

The robotic sorting system concept is an integrated solution for warehouse automation, built around a robotic arm with intelligent recognition and control capabilities. All system control, including vision processing and decision-making, is performed exclusively by Ambarella CV72 SoC under ROS2 without any additional MCUs and controllers. This ensures a compact architecture, reduced latency, and high energy efficiency without reliance on external compute units. The system is designed for fast, accurate, and continuous sorting of finished goods into predefined categories, storage zones, or shipping channels. This solution transforms warehouse sorting from manual or semi-automated processes into a fully digital, controlled environment where precision, speed, and predictability are key performance drivers.

This is AI Camera for industrial imaging from Rhonda Software, a compact and versatile robotic camera built on the Ambarella CV72S SoC with Robot Operating System and Ambarella Cooper SDK support, features a rugged industrial enclosure with a changeable C-mount lens.

Rhonda Software presents a super-compact reference design enabling embedded computer vision projects in constrained scenarios.
Leveraging the power-efficient Ambarella CV25 SoC, the Rhonda's system on module supports up to 4K video recording and streaming, addressing a wide range of applications for Edge AI.

The demo solution features the Rhonda AI Camera, a next-gen machine vision camera based on the Ambarella® CV72S silicon, a rugged enclosure, changeable C-mount lens, ROS2, and Ambarella Cooper SDK support, designed for use in changing production environments.

This demo camera setup scans PCBAs, instantly outlining good boards in white and highlighting defects in red directly on the operator's display.

Rhonda AI Camera, designed for industrial imaging - is a compact and versatile, self-contained vision system that integrates a camera, CVFlow® engine, lighting, and communication into one unit, enabling real-time image capture and analysis for automated tasks like inspection, measurement, and quality control in manufacturing, automating complex vision challenges without needing separate PCs.

Meet Rhonda Software FleetCam Proto - the Advanced Driver Assistance System (ADAS) and Driver Monitoring System (DMS) Platform based on the Ambarella CV25 SoC, with an integrated front-ADAS library developed by Brodmann17, and a DMS library provided by Jungo.

This Platform is capable of analyzing both on-the-road situations and driver attention, as well as issuing warnings and capturing corresponding video events remotely on Celesta ­- Rhonda Software's video cloud service. Cloud-based Web UI and Mobile App allow users to manage the connected cameras, and review on-the-road events gathered along with location data.

The Rhonda Software Fleet Cam Proto is ready for use and testing. It provides a solid basis for the development of customized Driver Safety Systems, that are in high demand in markets such as: commercial & passenger transport fleet management, telematics service providers, telematics insurance, car-sharing.

Rhonda Software is showcasing the Person Re-Identification demo, combined with Pose Estimation and Activity Recognition DNNs on a single Camera platform powered by an Ambarella CV2 SoC.
Our CV-team have designed and trained a bundle of new neural networks to enable the re-identification scenario with compact feature descriptors transferred between two camera units, so that a person visible on one camera will be successfully recognized on another.

The main difficulty in achieving this was in balancing performance and recognition quality to run all DNNs efficiently on the Rhonda Software CV2 System on a Module (SoM) platform.

As a result, the system is able to learn and re-identify the person in the live camera feed. The re-identification algorithm is sophisticated enough to do body tracking and appearance model re-learning on the go, when conditions allow. Identity models are shared with other cameras for multi-camera person tracking.
To demonstrate the power of DNN acceleration on the Ambarella CV2 platform in addition to re-ID, the Pose Estimation and Activity Recognition networks were running in parallel on each person in the camera’s field of view.

Rhonda Software helps customers to develop their camera products by providing turnkey solutions from the initial idea to the factory mass production. Also we can offer help with specific areas of camera development. We have extensive expertise in
- Still and Video processing,
- Connected devices,
- Computer Vision,
- Cloud based services,
- Mobile Applications, etc.

Rhonda Software at Embedded World 2020 showing 360 VR demo with real-time video stitching and live streaming over USB. This HW setup utilize two wide-angle lenses to capture panoramic video and perform image processing on board of H2 SoM - designed in house System on a Module based on Ambarella H2 SoC.

Rhonda Software implemented video streaming with WICED stack on RTOS. Camera application is based on Rhonda Camera SDK with Cypress WICED integrated on RTOS. Board's Wi-Fi AP is detected and connected in less than 4 seconds after power up.

Rhonda Software developed the Auto Exposure (AE) algorithm. Rhonda video decoder application receives video from the camera and allows user to select any region by touching the screen at an Android device. Selected coordinates are sent to the camera, and the camera applies window to the Rhonda AE algorithm. The algorithm, in its turn, calculates the best local exposure. Camera application is based on Rhonda Camera SDK.

Rhonda Software developed the Auto Focus (AF) algorithm. The camera streams video in 720p@30 H.264 over RTMP protocol with 1 Mbps. Rhonda video decoder application receives this video from the camera and allows user to select any region by touching the screen at an Android device. Selected coordinates are sent to the camera, and the camera applies window to the Rhonda AF algorithm. The algorithm, in its turn, finds the best focus position.

The Ambarella A9 platform makes the next step in dashcam technology. 4K is 8 megapixels, it's 4 times bigger than the commonly used Full HD resolution. This allows the recording of more fine details, such as plate numbers of distant cars. The up-to-date SDXC memory card standard is supported. A 64 GB card allows up to 3 hours of recording in 4K resolution. Do you want to publish or print moments from your recording? No problem. Snapshots extracted from 4K video are like 8 MP snapshots from photo cameras, so you can take any moment from the video and share or print it as a photo. Want to publish the action you captured? Even if it is far, zooming leaves enough resolution for good quality video.

Using Rhonda Camera SDK, Rhonda Software developed the application for dual video streaming from Ambarella H2 EVK Everest board with two Sony IMX377 image sensors 4K@30 60 Mbits (left sensor) + 4K@30 60 Mbits (right sensor) over 1 Gbit Ethernet.

Rhonda Software developed Rhonda Reference Design A12 micro board on the base of Ambarella A12 SoC. The reference design consists of the main board and sensor board. Rhonda Software implemented UVC 720p@30 video streaming from the micro board. Camera application of the micro board is based on Rhonda Camera SDK.

Rhonda Software implemented UVC (USB Video Class) 720p@30 video streaming to Raspberry Pi from Rhonda reference design board based on Ambarella A12 SoC, using Rhonda Camera SDK.

Rhonda Software implemented 1080p@30 5Mbps RTMP video streaming over Wi-Fi from Ambarella H22 EVK board. Latency is less than 300 msc. Camera application is based on Rhonda Camera SDK.

Rhonda Software have developed Rhonda Reference Design A12 micro board on the base of Ambarella A12 SoC, Murata Wi-Fi module and Sony image sensor. The reference design consists of the main board and sensor board. Rhonda Software implemented RTMP streaming over Wi-Fi from the micro board. Camera application of the micro board is based on Rhonda Camera SDK.

Rhonda Software implemented 1080p@30 10Mbps video streaming over Ethernet from Ambarella S5L Strawberry EVK board.

Rhonda Software implemented 720p@30 video streaming over Ethernet from two Ambarella EVK boards - Mango S3L and Olive S3Lm. Latency is about 300-330 msc.

Since 2009 Rhonda Software has been improving its proprietary barcode recognition algorithm, optimized for embedded platforms:
1) Barcode recognition - 2009
2) Barcode recognition (mobile platform) - 2009
It has been ported to Rhonda reference board based on Ambarella H2 SoC. The algorithm works fast, each barcode recognition takes less than a second. Barcode recognition application is based on Rhonda Camera SDK.

Rhonda Software implemented LCD support for Rhonda Reference Design A12 micro board based on Ambarella A12 SoC. The reference design consists of the main board and sensor board. Camera application of the micro board is based on Rhonda Camera SDK.

Rhonda Software developed Rhonda Reference Design A12 micro board on the base of Ambarella A12 SoC. The reference design consists of the main board and sensor board. Rhonda Software implemented UVC 720p@30 video streaming to Raspberry Pi from the micro board. Camera application of the micro board is based on Rhonda Camera SDK.

Rhonda Software has developed an application that allows an Ambarella CV22 SoC-based camera to be used to perform voice calls over a regular cell network (through Telit LTE module), with complete two-way sound reception and transfer. Camera application is based on Rhonda Camera SDK.

Rhonda Software developed a camera application that enables real-time video streaming from Ambarella CV22 EVK camera board over LTE connection to a cloud-enabled application installed on a given device. Camera is connected to LTE cellular network through a Telit LE910C1-EU LTE Modem evaluation kit. Camera application is based on Rhonda Camera SDK.

UVC streaming in 4Kp60 MJPEG and NV12 (Uncompressed) 720p30 with Rhonda H2 SoM