Detecting and classifying object images using ruby
summary
Ruby code uses pycall to run Python libraries so OpenCV image ops and deep models from Caffe execute inside a Ruby pipeline. Drone mosaics get tiled, vegetation‑index formulas extract bands, object proposals come from contours sliding windows and selective search, and models train with NVIDIA DIGITS in a Dockerized environment. The system produces chemical heat maps (example accuracy: 64% potassium, 74–75% phosphorus and organic matter), yields useful but sometimes hallucinating predictions, and requires iterative retraining and dataset cleanup.
AI-written summary · about
chapters (14)
- 0:00Object detection in Ruby
- 1:29Image processing research
- 3:37Image processing tools
- 6:11PyCall for Python
- 7:58OpenCV image loading
- 10:35Color channel analysis
- 12:44OpenCV object detection
- 14:56Deep learning classification
- 16:20Caffe for classification
- 19:16Custom model training
- 21:01Dissertation results
- 22:32Future steps
- 23:17Ruby in AI
- 24:23Q&A
related talks
31mExploring Real-time Computer Vision Using ActionCable
40mComputer Vision Using Ruby and libJIT
36mDeep Learning - Segmentation, Identification and Classification
51mMachine Learning with Ruby
29mA Magical Gathering
37mUnstructured Data Processing with a Raspberry Pi AI Kit and Python
31mFrom Radar Echoes To Real-Time Recognition: Deep Learning on Another Planet -An Open Source Workflow
41mKEYNOTE: Image processing, AI, and autonomous systems