<![CDATA[Fish object detection in digital images is an application of image processing that can support automatic object identification. This study aims to implement Contour Detection for detecting fish objects in digital images using Python and OpenCV. The processing pipeline consists of image acquisition, grayscale conversion, Gaussian Blur filtering, binary thresholding, contour extraction using findContours(), contour-area selection, and bounding-box visualization. Testing used 50 digital fish images with variations in object size, position, illumination, and background. The evaluation compared system predictions with the ground truth using a confusion matrix and calculated Accuracy, Precision, Recall, and F1-Score. The reported results were TP=41, TN=1, FP=6, and FN=2, producing Accuracy of 84%, Precision of 87.23%, Recall of 95.35%, and F1-Score of 91.11%. The results indicate that the applied Contour Detection pipeline can detect most fish objects, particularly when the contrast between the fish and background is sufficiently clear. Performance remains affected by uneven illumination, complex backgrounds, small objects, and partially visible fish.]]>
Copyrights © 2026
