DEV Community: NASEEM A P

Custom Human Pose Classification using Mediapipe

NASEEM A P — Mon, 26 Sep 2022 06:14:14 +0000

Creating a Custom pose classification using Mediapipe with help of OpenCV
GitHub: CustomPose-Classification-Mediapipe

Sample OutPut

Video

Image

Demo

I am going to create a Custom Human pose classification using Yoga Pose Dataset.

1. Clone the Repository:

git clone https://github.com/naseemap47/CustomPose-Classification-Mediapipe.git
cd CustomPose-Classification-Mediapipe

Install Dependency

pip3 install -r requirements.txt

Download Dataset:

wget -O yoga_poses.zip http://download.tensorflow.org/data/pose_classification/yoga_poses.zip

About Dataset:

5 Classes: Chair, Cobra, Dog, Tree and Warrior
Contain Train and Test data
Combine both Train and Test data

├── Dataset
│   ├── Chair
│   │   ├── 1.jpg
│   │   ├── 2.jpg
│   │   ├── ...
│   ├── Cobra
│   │   ├── 1.jpg
│   │   ├── 2.jpg
│   │   ├── ...
.   .
.   .

2. Create Landmark Dataset for each Classes

-i , --dataset __ Path to Dataset
-o , --save __ Path to save CSV file

python3 poseLandmark_csv.py -i data/ -o data.csv

CSV file will be saved in <path_to_save_csv>

3. Create DeepLearinng Model to predict Human Pose

-i , --dataset __ Path to CSV Data
-o , --save __ Path to save model.h5 file

python3 poseModel.py -i data.csv -o model.h5

Model will saved in <path_to_save_model> and Model Metrics saved in metrics.png

4. Inference

Show Predicted Pose Class on Test Image or Video or Web-cam

-m , --model __ Path to saved model.h5
-c , --conf __ Min prediction conf to detect pose class (0<conf<1)
-i , --source __ Path to Image or Video, for webcam use Zero(0)
--save __ It will save Images (on ImageOutput Dir) or Videos (“output.avi”)

# Image
python3 inference.py --model model.h5 --conf 0.75 --source data/test/image.jpg
# Video
python3 inference.py --model model.h5 --conf 0.75 --source data/test/video.mp4
# web-cam
python3 inference.py --model model.h5 --conf 0.75 --source 0
###### to save ######
# Image
python3 inference.py --model model.h5 --conf 0.75 --source data/test/image.jpg --save
# Video
python3 inference.py --model model.h5 --conf 0.75 --source data/test/video.mp4 --save
# web-cam
python3 inference.py --model model.h5 --conf 0.75 --source 0 --save

To Exit Window — Press Q-key

Custom Pose Classification

Clone this Repository:

git clone https://github.com/naseemap47/CustomPose-Classification-Mediapipe.git
cd CustomPose-Classification-Mediapipe
git checkout custom

1. Take your Custom Pose Dataset

Example:

├── Dataset
│   ├── Pose_1
│   │   ├── 1.jpg
│   │   ├── 2.jpg
│   │   ├── ...
│   ├── Pose_2
│   │   ├── 1.jpg
│   │   ├── 2.jpg
│   │   ├── ...
.   .
.   .

2. Create Landmark Dataset for each Classes

-i , --dataset __ Path to Dataset
-o , --save __ Path to save CSV file

python3 poseLandmark_csv.py -i <path_to_data_dir> -o <path_to_save_csv>

CSV file will be saved in <path_to_save_csv>

3. Create DeepLearinng Model to predict Human Pose

-i , --dataset __ Path to CSV Data
-o , --save __ Path to save model.h5 file

python3 poseModel.py -i <path_to_save_csv> -o <path_to_save_model>

Model will saved in <path_to_save_model> and Model Metrics saved in metrics.png

4. Inference

Open inference.py
change Line-43: According to your Class Names, Write Class Order

Show Predicted Pose Class on Test Image or Video or Web-cam

-m , --model __ Path to saved model.h5
-c , --conf __ Min prediction conf to detect pose class (0<conf<1)
-i , --source __ Path to Image or Video, for webcam use Zero(0)
--save __ It will save Images (on ImageOutput Dir) or Videos (“output.avi”)

python3 inference.py --model <path_to_model> --conf <model_prediction_confidence> --source <image or video or web-cam>
# to save
python3 inference.py --model <path_to_model> --conf <model_prediction_confidence> --source <image or video or web-cam> --save

Show Predicted Pose Class on Test Image or Video or Web-cam

To Exit Window — Press Q-key

Thank You…

Auto Annotation using ONNX and YOLOv7 model (Object Detection)

NASEEM A P — Mon, 26 Sep 2022 05:40:47 +0000

Annotation is very boring work, so I think that can we use our custom trained model (ONNX model) to annotate our new Data.

So I created a python module that can Auto-Annotate your Dataset using your ONNX mode.

I also added new Auto-Annotator using YOLOv7 model (.pb)

Link to GitHub Repository:-
https://github.com/naseemap47/autoAnnoter

We can convert other types of model Tensorflow(.pb) or PyTorch(.pth) or other Models into ONNX. That’s why I choose ONNX model to build my Auto-Annotator module.

Convert To ONNX Model

From Tensorflow (.pb) I hoping that, you are using Object Detection API for this.

https://tensorflow-object-detection-api-tutorial.readthedocs.io/en/latest/index.html

When the time of exporting, Use:-
--input_type image_tensor
NOT
--input_type float_image_tensor
Example:-

python .\exporter_main_v2.py --input_type image_tensor --pipeline_config_path .\models\my_ssd_resnet50_v1_fpn\pipeline.config --trained_checkpoint_dir .\models\my_ssd_resnet50_v1_fpn\ --output_directory .\exported-models\my_model

pb to ONNX
Follow tensorflow-onnx:-
https://github.com/onnx/tensorflow-onnx

pip install -U tf2onnx
python -m tf2onnx.convert --saved-model tensorflow-model-path --opset 16 --output model.onnx
default of 13 for the ONNX opset. If you need a newer opset, or want to limit your model to use an older opset then you can provide the --opset argument to the command. If you are unsure about which opset to use, refer to the ONNX operator documentation.

From Pytorch (.pth) [pth to ONNX] Checkout Links Below:

How to convert custom .pth Model to .onnx? · Issue #193 · danielgatis/rembg
Hey, first of all thanks for sharing your great work! As described in the docs I trained a custom model based on the…
github.com

(optional) Exporting a Model from PyTorch to ONNX and Running it using ONNX Runtime - PyTorch…
In this tutorial, we describe how to convert a model defined in PyTorch into the ONNX format and then run it with ONNX…
pytorch.org

Auto-Annotate — ONNX Model
Clone Git Repository:-

git clone https://github.com/naseemap47/autoAnnoter
cd autoAnnoter
Install Required Libraries:

pip3 install -r requirements.txt
Variables:

-x , --xml _____ for XML Annotations
-t, --txt _____ to annotate in (.txt) format
-i, --dataset _ path to Dataset
-c, --classes _ path to classes.txt file (names of object detection classes)
Example for classes.txt :

car
person
book
apple
mobile
bottle
....
-m, --model__ path to ONNX model
-s, --size__ Size of image used to train the your object detection model
-conf, --confidence __ Model detection Confidence (0<confidence<1)
For XML Annotations:

python3 autoAnnot.py -x -i -c -m -s -conf
For TXT Format Annotations:

python3 autoAnnot.py -t -i -c -m -s -conf
It will do all operation and the end of process you will get your Auto-Annotated data inside your data.
Auto-Annotated data will be with your corresponding Image data.
In this way you can easy check, the Annotation is Correct or NOT.

Auto-Annotate YOLOv7 Model
Clone Git Repository:-

git clone https://github.com/naseemap47/autoAnnoter
Variables:

-i, --dataset ____ path to Dataset
-m, --model ____ path to YOLOv7 model (.pt)
-c, --confidence ___ Model detection Confidence (0 python3 autoAnotYolov7.py -i -m -c
It will do all operation and the end of process you will get your Auto-Annotated data inside your data.
Auto-Annotated data will be with your corresponding Image data.
In this way you can easy check, the Annotation is Correct or NOT.

Auto-Annotated data accuracy is Completely depends on your Custom Model.
So, its better to check the Annotation is Correct or NOT.
Let me know your feedback about my Auto-Annotator
Thank you…

Auto Annotation using ONNX and YOLOv7 model (Object Detection)

NASEEM A P — Fri, 23 Sep 2022 04:33:39 +0000

Annotation is very boring work, so I think that can we use our custom trained model (ONNX model) to annotate our new Data.

So I created a python module that can Auto-Annotate your Dataset using your ONNX mode.

I also added new Auto-Annotator using YOLOv7 model (.pb)

Link to GitHub Repository:-

https://github.com/naseemap47/autoAnnoter

We can convert other types of model Tensorflow(.pb) or PyTorch(.pth) or other Models into ONNX. That’s why I choose ONNX model to build my Auto-Annotator module.

Convert To ONNX Model

1. From Tensorflow (.pb)

I hoping that, you are using Object Detection API for this.

https://tensorflow-object-detection-api-tutorial.readthedocs.io/en/latest/index.html

When the time of exporting,
Use:-
--input_type image_tensor
NOT
--input_type float_image_tensor
Example:-

python .\exporter_main_v2.py --input_type image_tensor --pipeline_config_path .\models\my_ssd_resnet50_v1_fpn\pipeline.config --trained_checkpoint_dir .\models\my_ssd_resnet50_v1_fpn\ --output_directory .\exported-models\my_model

pb to ONNX

Follow tensorflow-onnx:-

https://github.com/onnx/tensorflow-onnx

pip install -U tf2onnx

python -m tf2onnx.convert --saved-model tensorflow-model-path --opset 16 --output model.onnx

default of 13 for the ONNX opset. If you need a newer opset, or want to limit your model to use an older opset then you can provide the --opset argument to the command. If you are unsure about which opset to use, refer to the ONNX operator documentation.

2. From Pytorch (.pth) [pth to ONNX]

Checkout Links Below:

https://github.com/danielgatis/rembg/issues/193

https://pytorch.org/tutorials/advanced/super_resolution_with_onnxruntime.html

Auto-Annotate — ONNX Model

Clone Git Repository:-

git clone https://github.com/naseemap47/autoAnnoter
cd autoAnnoter

Install Required Libraries:

pip3 install -r requirements.txt

Variables:

-x , --xml _____ for XML Annotations
-t, --txt _____ to annotate in (.txt) format
-i, --dataset _ path to Dataset
-c, --classes _ path to classes.txt file (names of object detection classes)
Example for classes.txt :

car
person
book
apple
mobile
bottle
....

-m, --model__ path to ONNX model
-s, --size__ Size of image used to train the your object detection model
-conf, --confidence __ Model detection Confidence (0<confidence<1)

For XML Annotations:

python3 autoAnnot.py -x -i <PATH_TO_DATA> -c <PATH_TO_classes.txt> -m <ONNX_MODEL_PATH> -s <SIZE_OF_IMAGE_WHEN_TRAIN_YOUR_MODEL> -conf <MODEL_OBJCET_DETECTION_CONFIDENCE>

For TXT Format Annotations:

python3 autoAnnot.py -t -i <PATH_TO_DATA> -c <PATH_TO_classes.txt> -m <ONNX_MODEL_PATH> -s <SIZE_OF_IMAGE_WHEN_TRAIN_YOUR_MODEL> -conf <MODEL_OBJCET_DETECTION_CONFIDENCE>

It will do all operation and the end of process you will get your Auto-Annotated data inside your data.
Auto-Annotated data will be with your corresponding Image data.
In this way you can easy check, the Annotation is Correct or NOT.

Auto-Annotate YOLOv7 Model

Clone Git Repository:-

git clone https://github.com/naseemap47/autoAnnoter

Variables:

-i, --dataset ____ path to Dataset
-m, --model ____ path to YOLOv7 model (.pt)
-c, --confidence ___ Model detection Confidence (0<confidence<1)

python3 autoAnotYolov7.py -i <PATH_TO_DATA> -m <YOLOv7_MODEL_PATH> -c <MODEL_OBJCET_DETECTION_CONFIDENCE>

Auto-Annotated data accuracy is Completely depends on your Custom Model.

So, its better to check the Annotation is Correct or NOT.
Let me know your feedback about my Auto-Annotator
Thank you…