# [09-18 to 09-25]
## Task - 1: Improve performance of in-appropriate baby detector:

- Previous model was giving 81% accuracy on test set.
- With the new model I was able to get 90% accuracy on the same test set.

## Task - 2: Run in-appropriate classifier Inference on all video series data
- Done and uploaded to drive

## Task - 3: Write scripts to combine all metadata except face-id
- created template with day, video identity, frame id, body detection bbox, kid/adult label, face detection bbox, face orientation data, expression label, appropriate/not-appropriate label
- Done and uploaded to drive
- Useful for Analysis

## Next Tasks:
- Move on to Model Compression and finishing trying all the algorithms on BA/Customer classifier.