Java Based Object Recognition Application for Visually Impaired

Java Based Object Recognition Application for Visually Impaired

Prof. Sandeep Ganorkar1, Pratiksha Wadgure2, Aishwarya Dhapkas3,

 Khushbu Mahadule4, Modita Meshram5, Saurav Shende6

1. Professor, Information Technology, K.D.K. College of Engineering, RTMNU, Nagpur, Maharashtra, India

2. Student, Information Technology, K.D.K. College of Engineering, RTMNU, Nagpur, Maharashtra, India

3. Student, Information Technology, K.D.K. College of Engineering, RTMNU, Nagpur, Maharashtra, India

4. Student, Information Technology, K.D.K. College of Engineering, RTMNU, Nagpur, Maharashtra, India

5. Student, Information Technology, K.D.K. College of Engineering, RTMNU, Nagpur, Maharashtra, India

6. Student, Information Technology, K.D.K. College of Engineering, RTMNU, Nagpur, Maharashtra, India

7. Student, Information Technology, K.D.K. College of Engineering, RTMNU, Nagpur, Maharashtra, India

Abstract

It was difficult to detect items in real time and turn them into an auditory output. Many real-time object identification apps have been made possible by recent developments in computer vision. This essay explains a basic Android application. that would make it easier for those who are blind to comprehend their surroundings. Data about the surrounding area was recorded using the phone's camera.

where the object detection API of Tensor flow was used to perform real-time object recognition.

Android's text-to-speech feature was then used to turn the objects that were recognized into an audio output. Complex algorithm offline processing is now easy thanks to Tensor flow lite. It was discovered that the suggested system's overall accuracy was close to 90%.

Keywords- computer vision, machine learning, TensorFlow, TensorFlow Lite, object acknowledgment, and Android.