potac

6 ms·

1.
by potac 10mo ago
Not sure if I understood correctly. Can we as individuals contribute to any of these projects?
2.
by potac 11mo ago
Location: Spain/UK Remote: Yes Willing to relocate: Yes (Europe) Technologies: 3D reconstruction, neural rendering, multi-view geometry, Diffusion models, AR/VR, Python, PyTorch, OpenGL, C++ Résumé/CV: https://manu
3.
by potac 1y ago
I'm not saying the mayor should change what is not under his radar. I'm saying the mayor should attract private companies for locals that don't want to be public state workers.
4.
by potac 1y ago
Do you know the funny thing? There is a cycle path from Pontevedra that stops ~500 meters before Marin (although still Pontevedra jurisdiction). It has been like this for years. Those 500 meters are ocean. We can't cycle through...
5.
by potac 1y ago
It's true there is a bus to Marin every 20min but it uses a _single_ fixed route. People that live far away (>1km) from this path spend less time driving to Pontevedra than walking to the nearest bus stop. It is not just increasing
6.
by potac 1y ago
Huh! I didn't know this. Thanks for clarifying. Yeah, we all know what the Xunta is doing...
7.
by potac 1y ago
I'm from Pontevedra. It has been the major's long-term project (~ 20 years) to make the city for the pedestrians: and he's done it. This works mainly because of two things: 1) the city is small and it takes aprox 30 min to wa
8.
by potac 2y ago
I don't really get why they need the occlusion volume? What is its functionality? Doesn't rendering a pointcloud from a novel viewpoint already give the (dis-)occlusion mask?
9.
by potac 2y ago
Just look at Pontevedra in Spain. 90% pedestrianised city and several years with 0 traffic related deaths. Lovely city!
10.
by potac 3y ago
Thanks. What was confusing me is the kernel size 4. Normally in (2D) convolutions you have (in_channels, out_channels, k, k) for a kxk kernel size. In the example above it the k is the first dimension instead of the last. This is in PyTorch
11.
by potac 3y ago
Can anyone explain how conv works in that graph. You have a tensor of shape [2,4,16] and you convolve with a kernel of shape [4,16,8] and that gives you a [2,8] tensor? How's that possible?