AsposeAI class

The AsposeAI class is a powerful bridge between OCR and modern AI models. It connects LLM/VLM inference with recognition output, turning raw OCR text into cleaner, smarter, and more actionable results.

🚀 Why this matters

• Improve table reconstruction and layout quality with AI table correction.
• Fix OCR mistakes automatically with AI spell-check correction.
• Extract key phrases and numeric values with AI keyword detection.
• Bring your own model from Hugging Face and tune runtime settings for your workload.

🛠 Constructors

public AsposeAI();
public AsposeAI(ILogger logger);

You can also pass optional logging and customization callbacks.

Configuration

⚙️ AsposeAIModelConfig

Property	Type	Description
AllowAutoDownload	string	If true, the model will be automatically downloaded if not available locally.
DirectoryModelPath	string	Optional path where downloaded or processed models will be cached. If not set, a default system location will be used.
FileModelPath	string	Local path to the folder containing the model files. If specified, this will be used instead of downloading. Default empty.
HuggingFaceFileName	string	Optional model filename to download from Hugging Face (for example, TableGPT2-7B.Q4_K_S.gguf). Use it to pin an exact model artifact for reproducible results.
HuggingFaceRepoId	string	Hugging Face repository ID (for example, bartowski/Qwen2.5-3B-Instruct-GGUF). Set this to bring your own model and adapt AI behavior to your domain.
ContextSize	int	Maximum context window (tokens) for LLM/VLM inference. Increase it for harder documents and richer reasoning; reduce it to save memory. If null, the model default is used.
GpuLayers	int	Number of model layers to run on GPU. Increase for faster inference on supported hardware, or set to 0 for CPU-only execution. Default: 40.

🧠 AsposeAI Class – Core Methods

Method	Description
SetPostProcessor(IOcrAIPostProcessor processor, AsposeAIModelConfig config)	Adds a custom AI postprocessor to enhance OCR results.
RunPostprocessor(List<string> texts)	Enhances plain recognized text strings using registered AI modules.
RunPostprocessor(OcrOutput output)	Enhances structured OCR output using registered AI modules.
Dispose()	Cleans up and releases resources used by the AI component.
FreeResources()	Explicitly unloads AI models and clears memory.
ListLocal()	Lists all local AI models available in the configured folder.
GetLocalPath()	Returns the directory path of the current model location.
IsInitialized()	Checks if the AI engine and model are ready to use.