Last active
May 10, 2023 17:58
-
-
Save ankane/4a9681c8d9b9e814debe9e3ea836529d to your computer and use it in GitHub Desktop.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
require "onnxruntime" | |
require "mini_magick" | |
img = MiniMagick::Image.open("bears.jpg") | |
pixels = img.get_pixels | |
model = OnnxRuntime::Model.new("model.onnx") | |
result = model.predict({"inputs" => [pixels]}) | |
p result["num_detections"] | |
p result["detection_classes"] | |
coco_labels = { | |
23 => "bear", | |
88 => "teddy bear" | |
} | |
def draw_box(img, label, box) | |
width, height = img.dimensions | |
thickness = 2 | |
top = (box[0] * height).round - thickness | |
left = (box[1] * width).round - thickness | |
bottom = (box[2] * height).round + thickness | |
right = (box[3] * width).round + thickness | |
# draw box | |
img.combine_options do |c| | |
c.draw "rectangle #{left},#{top} #{right},#{bottom}" | |
c.fill "none" | |
c.stroke "red" | |
c.strokewidth thickness | |
end | |
# draw text | |
img.combine_options do |c| | |
c.draw "text #{left},#{top - 5} \"#{label}\"" | |
c.fill "red" | |
c.pointsize 18 | |
end | |
end | |
result["num_detections"].each_with_index do |n, idx| | |
n.to_i.times do |i| | |
label = result["detection_classes"][idx][i].to_i | |
label = coco_labels[label] || label | |
box = result["detection_boxes"][idx][i] | |
draw_box(img, label, box) | |
end | |
end | |
img.write("labeled.jpg") |
Hey @Ashvith, thanks for reporting! I've updated the gist and post for Ruby 3 and tf2onnx 1.9.
Hey @ankane , I was wondering if this Python code is unneeded, and is it possible to convert the model through Ruby?
python -m tf2onnx.convert --opset 10 --fold_const \
--saved-model ssd_mobilenet_v1_coco_2018_01_28/saved_model \
--output model.onnx
I'm not aware of any Ruby libraries that can convert TensorFlow models to ONNX.
Make sure you're using the latest version of tf2onnx to convert the model. Also, the blog post has info on how to check the input names.
great work
Where can I find tf2onnx.convert to create model.onnx?
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
This code snippet isn't working properly as of October 2021. It keeps throwing this error: