Object recognition technology has made considerable advancements across several fields, yet its benefits are mostly unavailable to visually impaired people. This study suggests an innovative way to close this gap by creating an item recognition system that is especially made for those who are visually impaired. Advanced machine learning methods are used in the suggested system, such as convolutional neural networks (CNNs) for emotion identification and threat detection, multi-task cascaded convolutional networks (MTCNN) for face detection, and single shot multibox detector (SSD) for object detection. With text-to-speech (TTS) technology, the system provides extensive real-time audio feedback by translating visual information into aural descriptions. With the help of this method, people may more successfully engage with their surroundings and develop critical environmental awareness. Extensive research assesses the system’s efficacy in detecting objects, identifying faces, identifying emotions, and issuing danger alerts. These studies also show that the system has the ability to greatly increase the freedom and accessibility of visually impaired people.