// the find
llSourcell/YOLO_Object_Detection
This is the code for "YOLO Object Detection" by Siraj Raval
A thin wrapper around thtrieu/darkflow that adds nothing beyond a YouTube tutorial notebook. It ports the original Darknet YOLO weights to TensorFlow 1.x, letting you run inference or fine-tune on custom datasets. The real repo is darkflow; this is just a companion to a 2017 video.
The darkflow core it wraps does something genuinely useful: it lets you load original Darknet .weights files directly into TF without manual conversion. The protobuf export path (.pb + .meta) is practical for deploying to mobile. JSON bounding box output is clean and easy to pipe into downstream code. The partial weight loading when retraining on fewer classes actually works well.
Frozen at TensorFlow 1.0 — nothing here runs on TF2, and the migration path is 'rewrite it yourself'. Last touched in 2020, YOLO has been through v3, v4, v5, v8, and v9 since then; the detection quality is generations behind. This repo is literally just the upstream darkflow with a notebook dropped on top — if you need darkflow, go to darkflow. There's no original work here to justify tracking this separately.