issues
search
rhysdg
/
vision-at-a-clip
Low-latency ONNX and TensorRT based zero-shot classification and detection with contrastive language-image pre-training based prompts
23
stars
1
forks
source link
Grounding DINO - ONNX inference class
#7
Closed
rhysdg
closed
3 months ago
rhysdg
commented
4 months ago
The feature is on it's way shortly but I have the model available
here
I'll take a look at dynamic quantisation after the class is built!
TensorRT viability checks shortly after calibration