There are lot of assumption which can lead to risks here.
1. We are assuming that each personnel will deliver the min level of quality as per the annotation guidelines.
2. Annotation fatigue and all are considered as 20% underutilisation but rather at times it could be higher
3. Redoing in case of mistakes and how to handle the rejected scenarios also needs to be considered if it is going to be 100% manual. 8 weeks has to consider all the quality check and final restructuring of the data in the shape required by ML team
There are lot of assumption which can lead to risks here.
1. We are assuming that each personnel will deliver the min level of quality as per the annotation guidelines.
2. Annotation fatigue and all are considered as 20% underutilisation but rather at times it could be higher
3. Redoing in case of mistakes and how to handle the rejected scenarios also needs to be considered if it is going to be 100% manual. 8 weeks has to consider all the quality check and final restructuring of the data in the shape required by ML team