NVIDIA / NVIDIA/TensorRT

Support lower fixed input resolution (e.g. 768×768) for Detectron2 Mask R‑CNN sample

Open
#4,670 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Module:Samples
Dominant language
C++
Stars
13.4k
Forks
2.4k
Avg merge
5d 3h
Merged PRs (30d)
2

Description

The samples/python/detectron2 sample currently only demonstrates using a fixed input size of 1344×1344 for exporting and running the Detectron2 Mask R‑CNN model. In many deployment scenarios (edge devices, real‑time applications), this resolution is quite high in terms of memory and latency cost.​

Would it be possible to officially support or document a smaller fixed input resolution, such as 768×768 (or other lower sizes satisfying the model/plugin constraints)? Even just a short note or example on how to safely change the input resolution in the sample (and what limitations to expect) would be very helpful.

Thanks!

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start in samples/python/detectron2 by comparing the current 1344×1344 export and inference setup with the requested 768×768 configuration. Verify whether the smaller fixed resolution satisfies the model and plugin constraints, then document or demonstrate the supported configuration and its limitations.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Feature
Difficulty
3/5
Estimated time
1-2 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.