Hi,
That’s not a question but image processing spam
First an issue
opened 11:44AM - 07 Apr 23 UTC
feature
category: dnn
### Descripe the feature and motivation
A novel model with published trained we… ights https://github.com/facebookresearch/segment-anything.
- [ ] Export model to ONNX
- [ ] Model can be imported with dnn
- [ ] Model can be inferred
- [ ] Compare accuracy with PyTorch
- [ ] Create a sample
### Additional context
_No response_
then google
and
The repository provides code for running inference with the SegmentAnything Model (SAM), links for downloading the trained model checkpoints, and example notebooks that show how to use the model. -...
finally
We introduce the Segment Anything (SA) project: a new task, model, and
dataset for image segmentation. Using our efficient model in a data collection
loop, we built the largest segmentation dataset to date (by far), with over 1
billion masks on 11M...
3 Likes
Oh – that’s pretty cool.
It seems like I should be able to get something like a semantic embedding vector for each segment out of this model (e.g. a descriptor of what each segment component is). I see
#1x256x64x64 tensor 0..255 segments??
SamPredictor.features = model.image_encoder(input_image)
Can this be used to cluster segments by their semantic content? Or track (find similar) segments between video frames?