Date: Oct 2023
Developed a native Android application using Kotlin and Jetpack Compose, integrating open-source on-device image classification and image description models. Leveraging the EfficientNet TensorFlow Lite model trained on the ImageNet dataset, the application achieved a 70% success rate in accurately classifying products. Additionally, the app incorporated an API-based pre-trained GPT-2 image captioning model, which aided in generating detailed descriptions for each product.