The demand for cloud computing systems has increased tremendously in the IT sector and various business applications due to their high computation and cost-effective solutions to various computing problems. This increased demand has raised several challenges such as load balancing and security in cloud systems. Numerous approaches have been presented for load balancing but providing security and maintaining integrity and privacy remains a less explored research area. Intrusion detection systems have emerged as a promising solution to predict attacks. In this work, we develop a deep learning-based scheme that contains data pre-processing, convolution operations, BiLSTM model, attention layer, and CRF modeling. The current study employs a machine learning-based approach to detect intrusions based on the attackers' historical behavior. Deep learning algorithms were used to extract features from the image and determine the significance of dense packets to generate the salient fine-grained feature that can be used to detect malicious traffic and presents the final classification using fused features.