Ship the AI
Leaf Identification using Deep Learning
A CNN-based leaf identification system with a Raspberry Pi camera at the edge and an AWS-hosted inference service in the cloud.
- client
- Academic Project
- when
- 2019-2020
- headline
- Edge + cloud CNN pipeline
01 · Problem
An early end-to-end edge + cloud ML pipeline: capture, inference in the cloud, answer back on the device. That is the same deployment shape used for applied AI today.
02 · What I built
- Collected and pre-processed a leaf image dataset with augmentation and normalization
- Trained a Convolutional Neural Network in Python for reliable leaf classification
- Integrated a Raspberry Pi + camera that sends live images to an AWS-hosted inference service
- Documented the workflow from data collection through training, deployment and on-device testing
03 · Result
- End-to-end edge + cloud computer vision pipeline
- Real-time species prediction returned to the device