Tencent / Tencent/ncnn

Model conversion issue SpaceToBatchND not supported yet

Open
#3,166 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
C++
Stars
23.8k
Forks
4.5k
Avg merge
2d 20h
Merged PRs (30d)
37

Description

error log | 日志或报错信息 | ログ

/tensorflow2ncnn model-frozen-1.pb model-frozen-1.param model-frozen-1.bin
SpaceToBatchND not supported yet !
nTblock_shape #3
Tpaddings #3
T #1
BatchToSpaceND not supported yet !
nT #1
Tcrops #3
Tblock_shape #3
SpaceToBatchND not supported yet !
nT #1
Tblock_shape #3
Tpaddings #3
BatchToSpaceND not supported yet !
nTcrops #3
T #1
Tblock_shape #3
SpaceToBatchND not supported yet !
nT #1
Tpaddings #3
Tblock_shape #3
BatchToSpaceND not supported yet !
nTcrops #3
T #1
Tblock_shape #3

model | 模型 | モデル

This model contains operations which are supported in Tensorflow, I know NCNN doesn't support these operations. Is there any alternate path i can take to make this work on ncnn. Does writing SpaceToBatchND as a custom layer in ncnn would help ? does customer layer run on GPU or DSP or does it fall back to CPU.

how to reproduce | 复现步骤 | 再現方法

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 with the tensorflow2ncnn model-frozen-1.pb conversion command and the reported SpaceToBatchND and BatchToSpaceND errors. Determine the supported conversion and custom-layer paths, including whether custom layers can use GPU, DSP, or only CPU execution. Done means a model containing these operations converts and runs successfully.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp, tensorflow
Domain
machine-learning
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.