Detection and Counting of Mango Fruits in Occluded Condition Using Image Analysis

In Agriculture, fruit farmers would be very helpful if they can monitor and estimate the yield before harvesting the fruits so that they can optimize and use the materials required more efficiently such as water consumption, fertilizers, and other agricultural chemical substances for every different location. This study proposed a method for detecting and counting the number of mangos in occluded conditions by evaluating the color filter and identifying the specific characteristics of fruit such as the homogeneity of fruit surface. This study fully made use of the information extracted from the created blobs after conducting histogram filtering such as blob weighting, evaluating the blob gradient topography and performing a hierarchical clustering. This method had a lower efficiency cost and did not need to determine the number of clusters to be searched. The function of this method was also improved by providing the information of the position and the number of fruits in the result images. This information could be used to make a precise detection. The images used in this experiment were 150 mango images divided into 30 training images and 120 testing images. The results of the experiments showed this method was able to detect mango, precision and false rates up to: 97.53%, 99.28%, and 0.72%, respectively. In general, the result of this study presented the total number of fruit detected by system of 646 images as the True Positive conditions from the total of 705 fruits, with an overall ratio of Recall, Precision, False rate of 91.63%; 97.88%, 2.12% respectively.