LLM: prompt configuration, event verification and frame comparison
A shared LLM guide for BOX 3.2 and BOX 5.0: scheduled classification, detection with bounding boxes, verification, additional context and frame comparison.
LLM · BOX 3.2 / BOX 5.0LLM documentation
Shared LLM guides for BOX 3.2 and BOX 5.0: prompt types, response formats, execution and troubleshooting.
Choose an LLM task
- Create a prompt and agree the JSON response
- Classify a frame on a schedule
- Find objects and return bounding boxes
- Have the model verify an event
- Add context to an event
- Compare several frames
- Check the model's response
- Diagnose a processing error
Connect Codex or ChatGPT through MCP
Connect an assistant to read permitted BOX data and prepare reports. MCP access is separate from configuring LLM prompts for video analytics.
LLM instructions
Shared guidance for BOX 3.2 and BOX 5.0: configure prompts and work with model responses.
Prompts and results
Analytics
LLM: model connections and prompt fields
LLM: model connections and prompt fields. Detailed instructions, field descriptions, procedures, and interface illustrations.
BOX 3.2 / BOX 5.0LLM: processing modes and schedules
LLM: processing modes and schedules: detailed instructions, field descriptions, procedures and interface illustrations.
BOX 3.2 / BOX 5.0Operations
LLM: testing and response errors
LLM: testing and response errors. Detailed instructions, field descriptions, procedures, and interface illustrations.
BOX 3.2 / BOX 5.0