In short: AI Product Detection recognises your specific products by their shape, colour, structure and packaging, so replicas are caught even when the image shows no logo or brand name.
Understanding AI Product Detection
Not every replica carries your logo. Many of the hardest to catch are unbranded copies of your most recognisable designs. AI Product Detection recognises the product itself from its physical features, so these posts are found even when there's no logo or brand name to search for.
It's trained on your actual product catalogue, down to individual models and SKUs, so it recognises your genuine products by their structure, colour, shape and packaging. That gives you an accurate view of which products are most at risk, so you can prioritise enforcement and protect revenue.
Note: AI Product Detection must be trained on your own catalogue, models and SKUs before it can recognise your products. To set it up for your Program, speak to your Account Manager.
1. What it detects
Because it looks at the product rather than text or logos, it flags both branded and unbranded infringements.
Minimalist items: recognises logo-free products from their distinct visual features and structure.
Unbranded replicas: catches counterfeits and lookalikes that text- or logo-based detection often misses.
Targeted product lines: shows which of your products are copied most often.
Less reliance on text: works independently of keywords and visible logos.
Learn more: Product Line & Products
2. How it works
The model learns what each of your genuine products looks like in its entirety, so it can spot a copy of it.
Instead of looking for explicit brand identifiers, it analyses the physical and structural attributes of the product in each image:
Structure: how the product is built and put together.
Shape: its overall outline and proportions.
Colour: its colourways and finishes.
Packaging: how it's boxed, wrapped or presented.
Because it knows the genuine product, it can spot a replica of a specific product line, even when the image contains no logos or brand names.
How it compares
Zeal has several AI detections. Each one answers a different question.
Detection | The question it answers | Example (Solène) |
AI Product Detection | Which of the brand's products is this? | An unbranded copy of a Solène tote bag |
AI Category Detection | What type of product is this? | Filed under Handbags › Tote Bags |
Logo Detection | Is the brand's logo in the image? | The Solène wordmark on a bag |
AI Image Feature Detection | Do the details point to a fake? | Uneven stitching or a cheap dust bag |
Learn more: AI Category Detection · Logo Detection · AI Image Feature Detection
3. Finding posts by product
Pull up every post where one of your products was recognised.
Once your model is trained and live, product detections appear in the Product filter.
Learn more: Product Filter
4. Using detections in Auto-Moderation
Let product detections do the first pass of moderation for you.
Product detections can drive your Auto-Moderation rules, either to flag potential design infringements or to classify posts by the specific product identified.
Learn more: Building Auto-Moderation Rules · Auto-Moderation
Quick reference
Not sure where to go next? Start here. Can't find what you're looking for? Use the search bar at the top of the Help Centre, or contact your Corsearch team.
If you want to know… | Look at… |
How to get AI Product Detection set up for your brand | Your Account Manager |
What kinds of infringement it catches | 1. What it detects |
What the model looks at | 2. How it works |
Which detection to use | How it compares |
Where a product was recognised | 3. Finding posts by product |
How to sort posts by product type | |
How to spot fakes by their details | |
How to find your logo in images |