idachallenge-baseline

KO | EN

gitlite — search

#ai #llm #go #machine-learning #rust #linux #nodejs #java #cli #kubernetes #ios #mcp #openai #open-source

idachallenge-baseline

(★ 9)

No description available.

idachallenge-baseline Latest Version Download

Download Latest Version (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Professional IDA Challenge Baseline 🏆
This repository provides the official baseline for the Professional IDA (Industrial Defect Analysis) Competition.

The IDA Competition contains two tracks:
- Track 1:Cross-Scenario Defect Detection

This track focuses on the adaptability of vision models. Participants must develop models that can
  accurately detect, localize, and classify defects across different, unseen production lines and varying
  imaging conditions.  
Refer to `Track1-CrossScenarioDefectDetection/Cross-Scenario Defect Detection Baseline.md`, 
this baseline trains an instance segmentation model（YOLOv8-seg） and evaluates its zero-shot/few-shot performance 
in other cross-domain regions.

- Track 2: Fine-Grained Severity Grading

This track focuses on the analytical depth of vision models. Moving beyond simple detection, 
participants must analyze detected defects to identify their specific mechanism and assign an industry-standard
  severity level (Acceptable, Marginal NG(Not Good), NG, or Gross NG).    
Refer to `Track2-FineGrainedServerityGrading/Fine-Grained Severity Grading Baseline.md`, 
this baseline starts with a YOLOv8-seg model for instance segmentation, and then calculates a severity score by 
applying a weighted combination of four metrics extracted from the masks: defect area, bounding box length/width, 
and mean grayscale contrast. Ultimately, this score is used to classify the final defect grade.

idachallenge-baseline

idachallenge-baseline Latest Version Download

Today's picks

Was this content helpful?

Recent Feedback

idachallenge-baseline

idachallenge-baseline Latest Version Download

Today's picks

Was this content helpful?

Recent Feedback

Download README