Cyanobacteria are a major cause of algal bloom, a rapid rise or overgrowth in the population of BGAs and algae in fresh or marine water systems. Gram-negative, oxygenic photosynthetic prokaryotes, and cyanobacteria (Blue Green Algae – (BGA)) are diverse in nature with a long evolutionary history. For the detection, monitoring, forecasting, and management of harmful algae populations, there is a need for rapid identification and classification of these cyanobacteria. Deep learning-based image processing methods can be the best way to extract qualitative information from microscopic images of BGA samples. In this study, the most prevalent classification model AlexNet and VGG16 were applied for features extraction and fusion was performed for the last fully connected layers. With these retrieved features, the proposed Alex-VGG convolution fusion network (CFN) was used for classification purposes. The suggested ensemble model outperforms classic CNNs in terms of accuracy rate (99.36 %) and ROC curve area (99 %) with a shorter execution time.