TrOCR / README.md
YakumoAkayi's picture
Agora board builds (Model Farm, verified on real boards)
ab235aa verified
|
Raw History Blame Contribute Delete
1.48 kB
metadata
license: other
license_name: aplux-model-farm
license_link: https://huggingface.co/APLUX-Agora/TrOCR/blob/main/LICENSE_MODEL_FARM.pdf
pipeline_tag: image-to-text
tags:
  - Agora
  - QNN
  - AIoT
  - Qualcomm

TrOCR model resources

用编码器和逐字解码器识别英文短文本,保留完整生成标记供导出。

Board builds that Agora runs and measures on real Qualcomm boards. Try it online, without a board: https://mega.aidlux.com/models/trocr

SoC · board Precisions Measured on Agora (default precision)
QCS8550 · 犀牛派 X1 (Rhino Pi X1) FP16 77.5 ms

Task: 文字识别. Inputs: image (image). Outputs: text (text), out (json).

Download

export AGORA_WORKSPACE="$HOME/my-trocr"
export AGORA_SOC='QCS8550'
hf download APLUX-Agora/TrOCR \
  --include "targets/$AGORA_SOC/*" --local-dir "$AGORA_WORKSPACE/resources"

Each targets/<SoC>/ holds model.json (default precision and variant paths) and one directory per precision. Context binaries are named model.bin; multi-file packages are kept as published by Model Farm.

Pre- and post-processing, the runner and an exported project: open the model on Agora and choose “部署指南”, or export any workflow that uses it.

Licence and provenance

Source model licence: see the original model's terms. Converted, deployable resources: APLUX Model Farm License (LICENSE_MODEL_FARM.pdf). See NOTICE.md and SHA256SUMS.