Turn abandoned patents
into your own filing
into your own filing
This invention describes a way to understand what's in an image by breaking it down into many overlapping sections. It then uses a multi-layered system to figure out the probability that each section belongs to a certain category, combining these probabilities to make an overall guess about what the image or its parts show. The claims specifically narrow this to forming partially coextensive hierarchical layered blocks from the image.
Why it matters: Filed before the deep learning revolution, the computational power and vast datasets available today make the described hierarchical, probabilistic image classification methods, particularly those using layered blocks, far more effective and feasible than in 2001. The methods described align with foundational concepts now greatly enhanced by modern AI.
AI gives you a few directions you could take this. Pick one, and we check whether your version is different enough to patent, then write the filing.
Reinvent this with AI
