Using TensorFlow with Go
summary
Shows that TensorFlow’s Go API can both train and serve real models using only Go tooling and containers. Implements a CAPTCHA solver by loading a provided model, using the CLI to inspect node names and shapes, and automating login with a cookie jar. Builds in-house face recognition on FaceNet (Sandberg) with Docker and Redis, starting from a 97% base model and achieving >80% accuracy with one example per identity and client-specific models retrained quickly.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 0:58Proof of Concept
- 2:29GitHub repo
- 8:25What is TensorFlow
- 10:20TensorFlow Tools
- 11:59TensorFlow Languages
- 12:25Inhouse Machine Learning
- 13:03Requirements
- 15:06Implementation
- 17:17Results
related talks
04mTensorFlow and Go
34mBuilding a Go AI with Kubernetes and TensorFlow
50mDeep Learning with Go
01h 18mPutting Tensorflow Models into Production
01h 18mPutting Tensor Flow Models Into Production
28m(Neural) Natural Language Processing in Go
41mComputer Vision with GO and OpenCV
37mSimplifying Training Deep & Serving Learning Models...using Tensorflow