A new method is proposed to detect the location of protozoa using edge extractor and Artificial Neural Networks (ANN). In this method, pre-processing is applied to the activated sludge images. ANN is classified with seven parameters that contain general features of protozoa. The features are obtained from properties of each region after applying edge extractor. Morphological operations are applied to obtain protozoa as a whole. A system has been developed to decide the regions of protozoa using the morphological procedure after ANN process. If the determined regions are not elliptical or circular areas, the regions are expanded according to the characteristics of them. Expansion direction is found by the orientation angle of the region. Accordingly, if it is required, the starting point can be changed. Then, active contour method is applied to the regions in which the exact location can not be determined. It is important to find the correct boundaries of the object and the first mask place should be chosen correctly. To do this, the convex area of the image is used. So, the initial mask is obtained automatically. Then, the obtained regions are evaluated again at the decision stage and protozoa are detected.