Comments (10)
If this is a question to install numpy in the environment of the DALI backend, then I think that it'd be a good idea for convenience. My original comment was about adding numpy as a dependency to DALI, which would not make sense.
from dali_backend.
There are several examples in the dali documentation where it uses numpy in them, the backend should support it
from dali_backend.
In those examples, numpy is used to produce some kind of data to be used together with the DALI pipeline. However, DALI doesn't need numpy to work and therefore, adding it to its dependencies would be in my opinion wrong.
from dali_backend.
What if you need to build a transformation matrix just like I'm showing in the example code? That's a very normal thing to do and I don't see the downside of adding numpy to the env
from dali_backend.
Hi @fran6co,
I'm sorry to disagree. We want to enforce as less dependencies over DALI as possible.
If we want to add all packages that DALI can cooperate with we would need to add a dependency to Torch and CuPy as DALI can use tensors from these frameworks.
from dali_backend.
I don't agree about that this is equivalent to CuPy or Torch. I created this issue not as just "DALI supports X then this backend should as well" but with a specific use case. Can you advice how would you solve it without numpy?
from dali_backend.
Can you advice how would you solve it without numpy?
Yes. You can achieve that like this:
identity_mat = fn.constant(shape=(3, 3), fdata=[1.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 1.0], dtype=types.FLOAT, device="cpu")
from dali_backend.
@jantonguirao thanks, I will try that
from dali_backend.
I agree, the PR is to add numpy to the backend unless I added it to the wrong place
from dali_backend.
The original issue here was to create an auxiliary matrix using numpy. @fran6co proposed to extend DALI Backend dependencies with numpy
. However, there is a solution to this issue without the need of introducing an extra dependency and therefore is preferable. Thus I'll close this issue.
Nevertheless, the question of adding more dependencies to DALI Backend is still a valid one. On one hand, there certainly are use cases, where such dependency would make development the inference solution way easier. On the other, such approach looks like a gift that will keep giving: there will always be an extra dependency to add. For the latter reason I believe that adding those in a way presented in #143 is not the wanted solution (although numpy is so popular, that we may end up with this). Instead, I believe there should be an easy-ish way to add any particular dependency to the deployed tritonserver
image. We'll put this to our backlog and we'll design something proper.
from dali_backend.
Related Issues (20)
- Can dali backend support default values or optional input? HOT 2
- Unexpected large memory needed for gpu resize HOT 4
- Error in thread 31: nvJPEG error (5): The user-provided allocator functions, for either memory allocation or for releasing the memory, returned a non-zero code. HOT 6
- Cannot compile dali_backend with older version of triton HOT 2
- how to provide batch input data for dali pipeline whicn input shapes [-1] HOT 1
- if I want to crop from different start point, how can I build pipe to do this? HOT 2
- Test issue
- Connecting InputOperator with no explicit inputs to Triton HOT 12
- Could not serialize dali.fn.python_function HOT 1
- when using crop_mirror_normalize func, Output layout "CHW" is slower than "HWC" HOT 5
- dlopen libcuda.so failed!. Please install GPU dirverTraceback (most recent call last): HOT 4
- Prefeed multiple input batches to the inference pipeline HOT 7
- Unable to load numpy module in a DALI backend HOT 3
- DALI pipeline in Triton - formatting InferInput batch of images for UINT8 HOT 3
- 'NoneType' object has no attribute 'loader' when trying to load DALI model. HOT 15
- How to format client code for inception example HOT 14
- How to get list of image paths into dali pipeline? HOT 4
- How to use scalar inputs HOT 3
- Video Input larger than max
- Missing conda env. in 24.04 breaks autoserialization
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from dali_backend.