How I created a synthetic COCO image dataset of yard weeds from scratch in a single day. I then trained a Mask R-CNN to perform image segmentation on …
9 Responses
Loved your story telling to start with. There are limited documentation on how to setup the annotation and creating datasets from scratch. Thanks for sharing.
I like that stuff. If you combine it with a database like they have in PlantSnap, you could have a blacklist of plants you dont want in your garden and automate everything else.
Awesome 👍. I am also very curious about AI ML weeding robot to enable new way of machine in Ecological Agriculture maybee. Making dataset is critical and your idea is very practical.
Great job….,👏👏 I know how hard it is to do as I have worked on a similar kind of project last month, but my project was to detect weeds in densely cultivated land.🙂
Loved your story telling to start with. There are limited documentation on how to setup the annotation and creating datasets from scratch. Thanks for sharing.
I like that stuff. If you combine it with a database like they have in PlantSnap, you could have a blacklist of plants you dont want in your garden and automate everything else.
Awesome 👍. I am also very curious about AI ML weeding robot to enable new way of machine in Ecological Agriculture maybee. Making dataset is critical and your idea is very practical.
You could have used VGG annotator.
Great job….,👏👏 I know how hard it is to do as I have worked on a similar kind of project last month, but my project was to detect weeds in densely cultivated land.🙂
You're in Austin too I can tell from your lawn. Mine's the same.
This really great content! Could apply this to items in supermarket? How is it different from ImageNET, which I'm pretty sure can run on an iPhone?
Very impressive. How many epochs did it train for? What was the accuracy score on the training dataset? on the unseen dataset?
roundup would kill your grass too. You need a broadleaf herbicide.