More often than not, sellers tend to underdeclare their parcel size in order to pay a lower fee or possibly due to the lack of measuring equipment at home. As a result, NinjaVan has to re-sort these parcels that have been flagged out as well as issue additional payment to these sellers which could lead to potential conflicts. Overall, this is very time consuming and resource-intensive.
Integrating computer vision to ensure accurate parcel sizing inputs from the seller's end to increase efficacy through the seller-to-warehouse process segment, omitting the 'rejection pile' of packages, as well as disputes between sellers and NinjaVan
- Users have to submit 2 pictures of their parcel (top view and side view) beside a bottle of water (reference object) into NinjaLens
- NinjaLens will evaluate the sum of sides and return the recommended size of parcel based on the metrics
Try out NinjaLens here -> https://marcussyeo-logistipals-main-vy9cvn.streamlit.app/
Done by:
- Marcus Yeo - Year 1 NUS Data Science & Analytics
- Kwang Yang Chia - Year 1 NUS Data Science & Analytics
- Anders Ooi - Year 1 NUS Business Analytics
- Sourick Paul - Year 1 NUS Business Analytics