TFOD-MASKRCNN - Tensorflow MaskRCNN With Python

Last update: Jan 18, 2022

Related tags

Overview

Tensorflow- MaskRCNN Steps

git clone https://github.com/amalaj7/TFOD-MASKRCNN.git

1.  conda create -n tfod python=3.6

2.  conda activate tfod

3.  pip install pillow lxml Cython contextlib2 jupyter matplotlib pandas opencv-python tensorflow==1.15.0 (for GPU- tensorflow-gpu)

4.  conda install -c anaconda protobuf

5.  go to project path 'models/research'

6.  protoc object_detection/protos/*.proto --python_out=.

7.  python setup.py install

Install COCO API

8) pip3 install "git+https://github.com/philferriere/cocoapi.git#egg=pycocotools&subdirectory=PythonAPI"

Resize images in a folder

9) python resize_images.py -d train_images/ -s 800 600

Put images and annotations in corresponding folders inside images/ (Annotations are in COCO format)

10)  python create_coco_tf_record.py --logtostderr --train_image_dir=images/train_images --test_image_dir=images/test_images --train_annotations_file=coco_annotations/train.json --test_annotations_file=coco_annotations/test.json --include_masks=True --output_dir=./

copy nets and deployment folder and export_inference_graph.py from slim folder and paste it in research folder

Training

Create a folder called "training" , inside training folder download your custom model from Model Zoo TF1 | Model Zoo TF2 , extract it and create a labelmap.pbtxt file(sample file is given in training folder) that contains the class labels
Alterations in the config file , copy the config file from object_detection/samples/config and paste it in training folder or else u can use the pipeline.config that comes while downloading the pretrained model
Edit line no 10 - Number of classes
Edit line no 128 - Path to model.ckpt file (downloaded model's file)
Edit line no 134 - Iteration
Edit line no 143 - path-to-train.record
Edit line no 145 and 161 - path-to-labelmap
Edit line no 159 - path to test.record

Train model

python train.py --logtostderr --train_dir=training/ --pipeline_config_path=training/mask_rcnn_resnet50_atrous_coco.config

Export Tensorflow Graph

python export_inference_graph.py --input_type image_tensor --pipeline_config_path training/mask_rcnn_resnet50_atrous_coco.config --trained_checkpoint_prefix training/model.ckpt-10000 --output_directory my_model_mask

Inference

Open object_detection_tutorial.ipynb and replace the necessary fields like model path, config path and test image path

Result

View tensorboard

tensorboard --logdir=training

Tensorflow2 - MASKRCNN Steps

Almost similar steps as above .

git clone https://github.com/tensorflow/models.git

cd models/research
# Compile protos.
protoc object_detection/protos/*.proto --python_out=.
# Install TensorFlow Object Detection API.
cp object_detection/packages/tf2/setup.py .
python -m pip install .

To test the installation

python object_detection/builders/model_builder_tf2_test.py

Then follow the above steps from 8 to 10 (includes downloading the pretrained model and editing the config file according to your needs)

Train the model

python model_main_tf2.py --pipeline_config_path=training/mask_rcnn_inception_resnet_v2_1024x1024_coco17_gpu-8.config --model_dir=training --alsologtostderr

View tensorboard

tensorboard --logdir=training

Export Tensorflow Graph

python exporter_main_v2.py \
    --trained_checkpoint_dir training/model_checkpoint \
    --output_directory final_model \
    --pipeline_config_path training/mask_rcnn_inception_resnet_v2_1024x1024_coco17_gpu-8.config

Inference

For TFOD2 , you can utilize inference_from_saved_model_tf2_colab.ipynb and replace the necessary fields like model path, config path and test image path

TFOD-MASKRCNN - Tensorflow MaskRCNN With Python

Related tags

Overview

Tensorflow- MaskRCNN Steps

Install COCO API

Resize images in a folder

Put images and annotations in corresponding folders inside images/ (Annotations are in COCO format)

Training

Train model

Export Tensorflow Graph

Inference

Result

View tensorboard

Tensorflow2 - MASKRCNN Steps

To test the installation

Train the model

View tensorboard

Export Tensorflow Graph

Inference

Owner

Amal Ajay

Code for paper " AdderNet: Do We Really Need Multiplications in Deep Learning?"

The fastest way to visualize GradCAM with your Keras models.

Summary of related papers on visual attention

Simple Pose: Rethinking and Improving a Bottom-up Approach for Multi-Person Pose Estimation

StackRec: Efficient Training of Very Deep Sequential Recommender Models by Iterative Stacking

Pure python implementation reverse-mode automatic differentiation

This is an official implementation for "AS-MLP: An Axial Shifted MLP Architecture for Vision".

TensorLight - A high-level framework for TensorFlow

The official codes for the ICCV2021 presentation "Uniformity in Heterogeneity: Diving Deep into Count Interval Partition for Crowd Counting"

Text to Image Generation with Semantic-Spatial Aware GAN

Mip-NeRF: A Multiscale Representation for Anti-Aliasing Neural Radiance Fields.

SparseInst: Sparse Instance Activation for Real-Time Instance Segmentation, CVPR 2022

Code for our ICCV 2021 Paper "OadTR: Online Action Detection with Transformers".

Examples of how to create colorful, annotated equations in Latex using Tikz.

Megaverse is a new 3D simulation platform for reinforcement learning and embodied AI research

Residual Dense Net De-Interlace Filter (RDNDIF)

VisionKG: Vision Knowledge Graph

[CVPR 2022] "The Principle of Diversity: Training Stronger Vision Transformers Calls for Reducing All Levels of Redundancy" by Tianlong Chen, Zhenyu Zhang, Yu Cheng, Ahmed Awadallah, Zhangyang Wang

Politecnico of Turin Thesis: "Implementation and Evaluation of an Educational Chatbot based on NLP Techniques"

Code Repo for the ACL21 paper "Common Sense Beyond English: Evaluating and Improving Multilingual LMs for Commonsense Reasoning"