BEFORE
AFTER
- The MVP can accurately count the seeds from the image input.
- Broken seeds are counted based on parameter user selection (area).
- The seeds need to be spread to improve the accuracy of the model.
The agricultural sector, particularly those engaged in crop production, often face significant challenges related to the efficient counting and quality control of seeds. These challenges can directly influence their operational effectiveness, production rate, and subsequently, their economic outcomes. The manual counting and quality inspection of seeds is time-consuming. Additionally, identifying broken seeds during the manual inspection process is critical because the germination rate of such seeds is typically lower, which impacts overall crop yield.
The development of an algorithm that accurately counts both entire and broken seeds would resolve these issues, improving efficiency and accuracy in seed counting, while also providing valuable quality control information that can enhance agricultural productivity.
The algorithm filters the input image, identifying and filling the contours by color contrast. The broken seeds are identified based on their area. The seeds need to be spread to enhance the accuracy.
The MVP was developed based on the quality inspection requirements of an Argentina company. This process allows automation of the counting of seeds and increases productivity.
CAPTIONS
BEFORE
AFTER