The client is a leading manufacturer of optical and photographic equipment, renowned for its state-of-the-art offerings. The company decided to integrate AI into one of its series of DSLR cameras as part of a business strategy of integrating high-end camera features into its consumer series.
To achieve this, the client reached out to HCLTech, which had been its innovation partner for 5 years. They wanted a sustainable roadmap for AI-driven innovations to stay ahead in this competitive landscape.
The Challenge
Cameras have constrained computing capability, memory and storage. Incorporating AI capabilities into such a device called for a highly optimized, custom-built engine that could be accommodated in the camera without compromising the performance.
It was a significant engineering effort that required specialized AI and embedded systems skills. The AI specialists at HCLTech handcrafted the neural networks end-to-end and successfully developed the AI model that could fit into the device and function in real time.
- Enabling seamless integration with the hardware while preserving computational efficiency
- Real-time processing and low power consumption, enhancing the user experience and fulfilling sustainability expectations
Developing the AI engine was only one aspect of the challenge; training it was another. Since cameras can be used to click photos in a variety of settings, the AI model needed to be trained in identifying the widest possible range of popular photography genres – food, portraits, pets, sports, wildlife, sceneries, etc. Millions of images were used to ensure exhaustive training of the AI engine.
Another major challenge was the client’s home country’s policies and regulations. Complete ownership of the data sets that were to be used for training the AI models was mandatory. We complied with every applicable rule without compromising on the quanta and diversity of the data needed for extensive and intensive training.
The Objective
The client wanted to broaden the market reach of its DSLR cameras beyond professional photographers to include beginners and casual users. However, traditional DSLR cameras offer extensive manual controls for setting white balance, ISO, aperture, shutter speed, etc. Such high levels of complexity and technicality make DSLR cameras overwhelming for photography enthusiasts.
This was a major roadblock for the company toward expanding into a new market segment. The client wanted to leverage AI to simplify the functionalities of the camera to provide a professional-grade photography experience regardless of the level of expertise of the user. In a nutshell, the client wanted to:
- Make professional-grade cameras easy to use for the masses
- Increase sales and expand user base
- Capture a new market segment
The Solution
To fulfill the client’s objectives, we developed and deployed software that added an AI layer between the user and the camera. It automatically configures the camera settings for non-professional photographers. Once the user selects the photographic mode, the AI system automatically enables relevant, recommended settings for that mode.
This includes features like:
- Auto-focus, auto-exposure, auto-white balance and scene composition
- Real-time camera settings guidance and automatic adjustment
The Impact
The AI-powered camera assistant has already been deployed and rolled out. Its simple interface appealed to a broader market, thus expanding the client’s market by creating a new segment – the prosumer market. By making DSLR cameras more accessible, our client positioned itself as an innovator in photography.
The highlights of the solution include but are not limited to:
- 99% reduction in AI model footprint
- 94% solution accuracy
- <10% false detection rate
- 30+ fps real-time performance
