Softnautics-Xilinx-OCR

Smart OCR solution using Xilinx Ultrascale+ and Vitis AI

AI Engineering, Platform Engineering /

The rich, precise high-level semantics embodied in the text helps understand the world around us and build autonomous-capable solutions that can be deployed in a live environment. Therefore, automatic text reading from natural environments, also known as scene text detection/recognition or PhotoOCR, has become an increasingly popular and important research topic in computer vision.

As the written form of human languages evolved, we developed thousands of unique font-families. When we add case (capitals/lower case/uni-case/small caps), skew (italic/roman), proportion (horizontal scale), weight, size-specific (display/text), swash, and serifization (serif/sans in super-families), the number grows in millions, and it makes text identification an exciting discipline for Machine Learning.

Xilinx as a choice for OCR solutions

Today, Xilinx powers 7 out of 10 new developments through its wide variety of powerful platforms and leads the FPGA-based system design trends. Softnautics chose Xilinx for implementing this solution because of the integrated Vitis™ AI stack and strong hardware capabilities.

Xilinx Vitis™ is a free and open-source development platform that packages hardware modules as software-callable functions and is compatible with standard development environments, tools, and open-source libraries. It automatically adapts software and algorithms to Xilinx hardware without the need for VHDL or Verilog expertise.

Selecting the right Xilinx Platform

The comprehensive and rich Xilinx toolset and ecosystem make prototyping a very predictable process expedites the development of the solutions to reduce overall development time by up to 70%.
Softnautics chose Xilinx Ultrascale+ platform as it offers the best of application processing and FPGA acceleration capabilities. It also provides impressive high-level synthesis capability resulting in 5x system-level performance per watt compared to earlier variants. It supports Xilinx Vitis AI that offers a wide range of capabilities to build AI inferencing using acceleration libraries.

Softnautics used Xilinx Vitis AI stack and acceleration utilizing the software to create a hybrid application and implemented LSTM functionality for effective sequence prediction by porting/migrating TensorFlow-lite to ARM. It is running on Processing Side (PS) using the N2Cube Software. Image pre- and post-processing was achieved using HLS through Vivado and Vitis was used for inferencing using CTPN (Connectionist Text Proposal Network). We eventually graduated the solution to real-time scene text detection with video pipeline and improved the model with a robust dataset.

Scene Text Detection

There are many implementations available, and new ones are being researched. Still, a series of grand challenges may still be encountered when detecting and recognizing text in the wild. The difficulties in natural scene mainly stem from three differences when compared to scripts in documents:

  • Diversity and Variability are arising from languages, colors, fonts, sizes, orientations, etc.
  • Vibrant background on which text is written
  • The aspect ratios and layouts of scene text may vary significantly

This type of solution has extensive applicability in various fields requiring real-time text detection on a video stream with higher accuracy and quick recognition. Few of these application areas are:

  • Parking validation — Cities and towns are using mobile OCR to validate if cars are parked according to city regulations automatically. Parking inspectors can use a mobile device with OCR to scan license plates of vehicles and check with an online database to see if they are permitted to park.
  • Mobile document scanning — A variety of mobile applications allow users to take a photo of a document and convert it to text. This OCR task is more challenging than traditional document scanners because photos have unpredictable image angles, lighting conditions, and text quality.
  • Digital asset management – The software helps organize rich media assets such as images, videos, and animations. A key aspect of DAM systems is the search-ability of rich media. By running OCR on uploaded images and video frames, DAM can make rich media searchable and enrich it with meaningful tags.

Softnautics team has been working on Xilinx FPGA based solutions that require design and software framework implementation. Our vast experience with Xilinx and understanding of intricacies ensured we took this solution from conceptualization to proof-of-concept within 4 weeks. Using our end-to-end solution building expertise, you can visualize your ideas with the fastest concept realization service on Xilinx Platforms and achieve greatly reduced time-to-market.

Read our success stories related to Machine Learning expertise to know more about our services for accelerated AI solutions.

Contact us at business@softnautics.com for any queries related to your solution or for consultancy.

Source: Xilinx

Smart OCR solution using Xilinx Ultrascale+ and Vitis AI Read More »

fpga-market-trends-with-next-gen-technology

FPGA Based Solutions With Evolving Technologies

Platform Engineering, VLSI Design & Verification /

Propelled into existence riding on low NRE (Non-recurring Engineering) cost, the FPGAs began as an alternative to ASIC until the customer required beyond a number of units, which was called a cross-over point at which higher NRE cost of ASIC was justifiable. Slowly the flexibility of programmable logic helped FPGA vendors create a room for their product. The IEEE article mentions that 2/3rd of the designs were losing money at one point because of changing requirements, product failures, or outright design errors.

Today, FPGAs are forming the backbone for 5G, Embedded Vision, Smart World (Cities, Factories, and so on), Cloud platforms, and safety-critical systems. FPGAs are finding application in various sectors which have been very specific about performance and compliance requirements such as defense, aerospace, and automotive.

The differentiator for FPGAs has been the flexibility and the operating range. On the one hand, FPGAs can power the high-performance cloud data-centers requiring as much as several hundred watts, and on the other hand, powering feature-light apps running low-power designs drawing 1/1000th of a watt (1mW). Thus, FPGAs can accelerate searches for Bing search engine on Microsoft at lower power consumption. And the same assembly could also be hosting specialized low-power FPGAs to run specific operations such as controlling the system, securing firmware, etc.

Machine Learning and Artificial adoption in our world is going to boost the demand for FPGAs, considering these fields are still evolving. They need flexible programmability to support agile development cycles of end use-cases. There are so many cases where low-power FPGA designs can support object detection, counting operations, key phrase detection to enable complex use-cases, which make a more durable case for mass-adoption than ever foreseeable. However, it requires FPGAs to support stringent low-power demands at much smaller form factor.

Lattice Semiconductor provides specialized low-power FPGA offering to perform computer vision and AI inference applications. Adopting a platform-based approach to product development resulting in maximum design-reuse, Lattice Semiconductor has launched platforms more frequently at a considerably lower cost. Lattice Nexus and Lattice Crosslink-NX platforms are likely to extend the low-power advantage that earlier FPGAs have enjoyed for a long time.

The Edge computing devices, a backbone of ML/AI advancements, have been constrained by battery juice and connectivity speed for fast and frequent data transmission. The AI has seen limited adoption mainly because of inadequate ability to process large datasets caused by the low transmission per set timeframe. However, with 5G adoption around the corner, this limitation will become a thing of the past. Now, with both problems, low-power and transmission speed, being addressed, these FPGA based solutions are all set to reach mass-production as more connected devices proliferate riding on the widespread 5G network.

The low-power FPGAs providers are going to see consistently higher demand for FPGA units. However, they are going to be continuously challenged to serve the incoming custom-design requirements considering large deviations in end-use. That’s where the boutique design houses can offload the FPGA companies and become the enablers for the mass-adoption of AI applications running on FPGAs. Someone who can handle RTL complexities, build required ML firmware, edit drivers, and get the system working for the desired ML use case.

Know how Softnautics can help you design FPGA-Powered ML solution for your use-case.

FPGA Based Solutions With Evolving Technologies Read More »

Tracking Social Distance What Enables Those Red and Green Boxes

Tracking Social Distance: What Enables Those Red and Green Boxes

Platform Engineering /

Xilinx research shows its Ultrascale+TM XCVU13P FPGA (38.3 INT8 TOP/s) platform provides the same computing power as Tesla P40 (40 INT8 TOP/s) but flexibility and on-chip memory on Xilinx device results in significantly higher computing capability for different workloads and apps. Another Xilinx research in a general-purpose compute efficiency shows its Virtex Ultrascale+ performing 4x better than Nvidia Tesla V100. The scales tilt further in favor of FPGAs when we consider functional safety demanded by safety-critical aviation, autonomous automotive, and defense applications, such as ADAS.

Embedded Vision requires the machines which have the ability to see, sense, and quickly respond to challenges the hardware designers create next-gen architecture that is highly differentiated and extremely responsive to adapt to ever-evolving algorithms and image sensors.

If the ones mentioned above are computing power-intensive use-cases that need to utilize custom neural networks (CNNs/DNNs), the other side of the spectrum requires extremely low-power operations with a flexible solution building approach.

In pursuit of maximizing the efficiency of machines to achieve higher throughput of operations, the Industrial Internet of Things (IIoT) is driving Industry 4.0. Such applications require a combination of software programmability coupled with real-time processing of sensor data to leverage any-to-any connectivity in a secure and safe manner. The flexible nature of FPGA programmability and low-power consumption make FPGAs a perfect choice for Industry 4.0 solutions.

It is just the beginning of how FPGA platforms are powering the ML solutions that are likely to see mass-adoption and become household essentials as we create diverse use-cases to help people, businesses, and governments to make the world a safer and smarter place
Know how Softnautics can help you design FPGA-Powered ML solution for your use-case.

Tracking Social Distance: What Enables Those Red and Green Boxes Read More »

3 Reasons Why Your IoT Mission Has Been Failing To Reach Autopilot

3 Reasons Why Your IoT Mission Has Been Failing To Reach Autopilot

Digital Engineering /

Only 1 in 4 companies drive business value from their IoT mission, the rest 75% are still figuring out a way to extract business value from the technology investments. The IoT as technology has evolved in multiple directions and enterprises constantly struggle in deciding whether they are choosing the right set of components. It can be attributed to the absence of a governing body and a lack of standardization resulting in interoperability issues during technology evolution. If you feel nothing has happened in your company’s IoT mission over the last 2 years, this blog can ring some bells on possible causes.

Reason 1: Only the latest technology can solve modern challenges
The beauty of IoT lies in the fact that it does not necessitates the need to use the latest technology as it works on the concept of a good-enough solution to fix the problem at hand. It means you shouldn’t get subsumed by overwhelming jargon floating around the world of IoT from devices to software to apps.

Reason 2: Our problem is very sophisticated, hence require complex solution
The way most problems are formulated, they appear very sophisticated and demand cutting-edge solutions. Seasoned technology buyers understand this nature, but they face challenging questions from requesting organizations about why they are not going for the latest tech available in the market. A good understanding of the problem boundaries and the proposed solution will equip these buyers to make a convincing case to go for a good-enough solution instead of chasing the excellent tech-specification resulting in over-solving the problem.

Reason 3: Latest technology will be long-lasting
This is the most common myth that has been propagated over the years. With passing time, the technology cycle has shortened, and businesses are not regularly facing the challenge to upgrade their service model. It means, neither the latest technology will necessarily last for many years, nor your business will need an over-sophisticated solution. As a result, the focus needs to be on managing cost instead of pondering over feature-richness and longevity of the latest tech-solution.

The IoT is a lot like Mathematics, multiple ways to solve a problem and reach the same solution. It means, businesses need to define the problem properly and then trust the solution provider to solve the problem instead of breaking head over finding the best-component and finding the latest-solution. It is partly because of the evolving nature of business models due to constant disruption and challenges posed by disruptions from unexpected quarters. When you don’t know how you will serve your customers 2-years down the lane, why bother about creating a long-lasting IoT solution to solving today’s challenge?

Hence, IoT solutions are like a rolling investment where business owners revisit your business model periodically to re-define problem statement which is relevant to the prevailing scenario in the competitive landscape. And the fact that your business is going to evolve every quarter to respond to upcoming disruptions, soon your problem statement will change, and you will need the right counsellor to guide you with hardware specs and solution changes.

Know about the Softnautics services to overcome your challenges!

3 Reasons Why Your IoT Mission Has Been Failing To Reach Autopilot Read More »

Scroll to Top