Netra is a startup developing Image Recognition technology based on AI / Deep Learning research conducted at MIT's Computer Science and Artificial Intelligence Laboratory. Netra's software brings structure to previously unstructured imagery with some amazing clarity. Within 400 milliseconds, Netra can tag a scanned image for brand logos, image context, and human face characteristics.
Consumers share 3.5 billion photos on social media every day. Within socially shared imagery are valuable insights about consumers' activities, interests, brand preferences, relationships, and key life events.
At Netra, we use AI, computer vision, and deep learning to help marketers better understand what consumers are already sharing; our technology can read images on a massive scale not previously possible. In order to accomplish this, we start with a sampling of images found online that contain a particular logo. We then take, say, a Starbucks logo, and alter it several different ways to create a training set that will allow the tech to recognize Starbucks logos that are distorted, or in crowded scenes like a coffee shop. Then we train the computer models using a combination of the organic content and synthetically altered images. Richard Lee, CEO, Netra
Below is an example of an image that Netra software ingested from Tumblr. Even though the caption does not mention The North Face, Netra’s software is able to scan the photo and detect the presence of the logo among other items of interest, including:
- Objects, scenes, and activities like Mountaineering, Summit, Adventure, Snow, and Winter
- A white male aged 30-39
- The North Face brand logo with 99% confidence
Netra offers customers access to a web-based dashboard to upload imagery and/or analyze social imagery sourced from Twitter, Tumblr, Pinterest, and Instagram. The software is commercially available for customers via a web-based dashboard or API for enterprise software companies. Netra’s core technology can also be applied including image indexing and search (digital asset management) and visual search.
Users can view analytics on the image tags and answer key questions such as:
- Where is my brand showing up in imagery and in what context?
- What demographics are engaging with my brand in imagery?
- What demographics are engaging with my competitors’ brands?
- What activities/brands are consumers that engage with my brand also interested in?
Users can filter imagery based on engagement levels as well as the context of the photo. Netra also has the ability to create a custom audience based on the content posted within social media imagery. For example, Reebok could utilize the software to target consumers who actively workout by targeted Crossfit to consumers who have posted photos of themselves engaged in exercise activity in the last two weeks.
We believe we have best-in-class tech in the brand and logo detection market. We also differentiate ourselves with additional image recognition capabilities. There’s only one other company that can do brands, logos, objects, scenes, and humans, and that's Google. In our head to head tests, we perform two times better than them. Netra’s visual intelligence solution can provide incredibly valuable data to augment existing consumer data (e.g. profile information, text captions, cookie data) that social advertisers already leverage. Richard Lee, CEO, Netra
Practical applications include brand monitoring, social listening, social advocacy, influencer marketing, marketing research and advertising.