Data validation and processing services
Summary
Validation of collected datasets (e.g., check for presence of all datastreams, check for drop-outs, check for focus or exposure of camera images). Preparation for performance evaluation (e.g., formatting; identification of the images containing selected visual markers; association between images from the system under test and the ground truth).
S00065
Version: 1.10
- arable farming
- horticulture
- greenhouse
- tree crops
- viticulture
- livestock farming
- food processing
- desk assessment
- test design
- test setup
- provision of datasets
- ELSA assessment
- LCA assessment
- conformity assessment
- AI model training
- people training
- physical system
- software or AI model
- data
- design / documentation
- other (including: nothing)
- Location: in Poland
- Offered by: L-PIT
Description
Data validation ensures quality before training or evaluating AI systems. It involves checking for completeness, like missing sensor data or camera images. Continuity is also crucial, identifying and potentially fixing gaps like missing timestamps or dropped video frames. For image data, focus and exposure are vital, as blurry or poorly lit images can hinder performance. Techniques like interpolation, imputation, sharpening, or filtering might be used to address these data quality issues. Once validated, the data is prepared for performance evaluation.