Automated image captioning system with deep learning enabled optimized approach

Image caption-generating systems aim to deliver accurate, coherent, and useful captions. This includes identifying the scene, items, relationships, and attributes of the image's objects. Due to constraints in using all visual information, captioning images can be complex. The proposed Hybrid Chimp Wolf Pack Inception-V3 (HCWPI) -Bidirectional Gated Recurrent Unit (BiGRU) approach uses encoding and decoding units to produce accurate captions for inputted emotion based images. The integration Hybrid Chimp Wolf Pack (HCWP)with Inception-V3 demonstrates a novel approach in encoding part. Initially, the Hybrid, Chip Optimization algorithm (COA) with Wolf Pack Optimization (WPO) namely HCWP incorporated with Inception-V3 is employed to generate appropriate input image reconstructions with fixed-length vectors, displaying the distinctive features of what was accomplished at the stage of encoding. The hyper parameter optimizer of HCWP optimizes Inception-V3 model performance. Descriptive phrases originate via a Bidirectional Gated Recurrent Unit (BiGRU) framework provided within the decoding part. Simulations were run to assess the HCWPI-BiGRU model's improved performance from multiple angles. The experimental results showed that the HCWPI-BiGRU model outperformed than existing approaches on term of blue and Meteor Score such as 0.86and 0.85 respectively.