* Gemm and MatMul block quantization support
* refactoring
* fix indentation
* node name independent
* Block quantization tool:
- constant weight category supported
- add data type saturation
- handled the case in which all the elements within a block are the same
benchmark script modified to support block quantized models
block quantized some models
* add missing block quantized models
* formatting
* add blocked models to eval script. Evaluation yunet
* Add sface and pphumanseg evaluation, block quantization tool fix, handpose blocked model fix, removed blocked CRNN EN,
* changed evaluation metric in block_quantize script and add verbose mode
* Add evaluation for PP-ResNet and Mobilenet
* changed file suffix and update readmes
* renamed int8bq
[GSoC] Gemm and MatMul block quantization support (#268)
* bump version to 4.8.0
* update benchmark results on i7 & rpi
* correct python command on i7
* update results on jetson nano
* update results on atlas 200 cpu
* update results on vim3 cpu, sunrise x3 cpu
* update results on vim3 npu
* update results on edge2
* update results on rv1126 and vision2
* update results on axp
* update table
* The results of the two models remain the same accuracy.
* re-upload int8 models
* updated by yiyao
* remove the sim suffix
---------
Co-authored-by: zihaomu <zihaomu@outlook.com>
Update handpose estimation model from MediaPipe (2023feb) (#133)
Wanli
50fc340
* update handpose model
* update quantize model
* fix quantize path
* update readme of quantization and benchmark result
* fix document
move setting description outside for loop in evaluation scripts (#134)
Labib Asari
5de3983
added evaluation script for PPHumanSeg model (#130)
Labib Asari
12c3d6b
* added evaluation script for PPHumanSeg
* added quantized model, renamed dataset
* minor spacing changes
* moved _all variables outside loop and updated accuracy
* removed printing for class accuracy and IoU
* added 2 transforms
* evaluation done on same size tensor as input size with mIoU 0.9085
* final changes
* added mIoU and reference
Update palm detection model from MediaPipe (2023feb) (#128)
Wanli
260eb6d
* update to the latest palm detection model
* quantize palm detection model
* make palm aligned to center when resize
* update benchmark for palm detection
* update benchmark data
* update readme
* update a new quantized model
* update readme
Update FER quantized model and fix document (#129)
Wanli
ebeb80f
* update document and fix bugs
* re-quantized model to per_tensor mode
* update KV3-NPU benchmark result
removed extra key in models.crnn dictionary (#126)
labeeb-7z
aa30eea
Add model file collector, fix some bugs and add some features (#123)
add model path searcher and remove hardwritten file paths from benchmark configs; pack ppresnet, mobilenet & crnn with labels; fix palm det data; add flags to enable models of different precision separately (#123)
Add a model for facial expression recognition (#100)
Chengrui Wang
596a24b
Re-quantify some models from per_channel mode to per_tensor mode (#90)
Wanli
323da84
* re-quantize some models from per_channel mode to per_tensor mode
* remove the description about per_channel
fix some error and bugs (#112)
Wanli
90edc6d
update benchmark (#97)
Wanli
8b5437e
revise the quantize-ort to quantize crnn into int8 with high accuracy (#94)
Yiyao Wang
d30c3db
* revise the quantize-ort to quantize crnn into int8 with high accuracy
* change the comment to English
* reduce supported suffix
Text Recognition: Add script to evaluate text recognition by ICDAR2003 (#71)
Yiyao Wang
46b1f95
* update readme
* add another script
* revise details for this pr
Add hand pose estimation model from Mediapipe (#83)
Wanli
57699b7
use QuantFormat.QOperator by default to avoid fake quantization (#88)
* replace with yunet of fixed input shape
* update quantized yunet
* update yunet filename used in scripts
* add a note message for https://github.com/opencv/opencv_zoo/issues/44
Add tools for quantization and quantized models (#36)
Yuantao Feng
55654c5
* add scripts for quantization
* update path to pp-resnet50
* add quantized models
* rename dict to models
* add requirements and readme
* fix typos
Shuyao "Tim" Xu +1: Tim-Xu * Add preprocessing with optimization before quantization * Clean up unintended changed --------- Co-authored-by: Tim-Siu <shuyao@u.nus.edu>2129055