dnth

5 ms·

1.
Show HN: X.infer-Framework agnostic computer vision inference(github.com)
3 points by dnth 2y ago|0 comments
2.
Show HN: Bag of Tricks to 8x Faster Inference with ONNX+TensorRT(dicksonneoh.com)
3 points by dnth 2y ago|0 comments
3.
Visualize your dataset using DINOv2 embedding
1 points by dnth 3y ago|0 comments
4.
by dnth 3y ago
We released a free tool (fastdup) where you can use to cluster image using DINOv2 embeddings.
5.
Cluster images at scale using DINOv2 embeddings(nbviewer.org)
1 points by dnth 3y ago|1 comments
6.
Find image duplicates and outliers – A free, scalable, efficient tool(github.com)
2 points by dnth 4y ago|1 comments
7.
by dnth 4y ago
fastdup is a tool that let you gain insights from a large image/video collection. It lets you identify image duplicates, video duplicates, wrong labels, outliers, corrupted data, and image clusters. fastdup is - Unsupervised: fits any
8.
Dedup-ing LAION (60M duplicates) and ImageNet (1.2M duplicates) with fastdup(youtube.com)
3 points by dnth 4y ago|1 comments
9.
by dnth 4y ago
The authors at fastdup ran an analysis on LAION 400M and Imagenet21K. Here's what they found. LAION 400M > 60M duplicates. > 962K broken images. > Various label discrepancies. ImageNet21K > 1.2M duplicate images. > 104K
10.
by dnth 4y ago
Even pros have dark, blurry & duplicate shots. But disorganization can make it hard to find those special memories. Let's fix that.
11.
by dnth 4y ago
Update - After some code optimization, I got the inference time down to below 100ms. The lowest I got on my Pixel 3 XL is 37ms! https://dicksonneoh.com/portfolio/pytorch_at_the_edge_timm_t...
12.
by dnth 4y ago
You're welcome! I'm not sure if I'm the right person to advise on this. But this idea is also known as federated learning right?
13.
by dnth 4y ago
With flutter you can also build web apps. https://flutter.dev/multi-platform/web
14.
by dnth 4y ago
Note yet! But I heard things are going to be a lot faster in 2.0. Have you tried?
15.
by dnth 4y ago
Thank you for the feedback! Let me know if you have questions :)
16.
by dnth 4y ago
Thank you!
17.
by dnth 4y ago
Thank you!!!
18.
by dnth 4y ago
Thanks a lot there! I was hesitating to write this piece actually, thinking it's not going to be valuable. I'm glad you find value in them!
19.
by dnth 4y ago
Media pipe looks really cool. I havent tried it. Have you?
20.
by dnth 4y ago
The pytorch lite package also supports yolov5 models. I posted on my LinkedIn awhile ago https://www.linkedin.com/posts/dickson-neoh_deploying-object...
21.
PyTorch at the Edge: Deploy 964 TIMM Models on Android with TorchScript(dicksonneoh.com)
102 points by dnth 4y ago|34 comments
22.
by dnth 4y ago
Model deployment is painful. Running a model on a mobile phone? Forget it . The frustration is real. I remember spending nights exporting models into ONNX and it still failed me. Deploying models on mobile for edge inference used to be com