Official code for ROCA: Robust CAD Model Retrieval and Alignment from a Single Image (CVPR 2022)

Related tags

Computer VisionROCA
Overview

ROCA: Robust CAD Model Alignment and Retrieval from a Single Image (CVPR 2022)

Code release of our paper ROCA. Check out our video, paper, and website!

If you find our paper or this repository helpful, please cite:

@article{gumeli2022roca,
  title={ROCA: Robust CAD Model Retrieval and Alignment from a Single Image},
  author={G{\"u}meli, Can and Dai, Angela and Nie{\ss}ner, Matthias},
  booktitle={Proc. Computer Vision and Pattern Recognition (CVPR), IEEE},
  year={2022}
}

Development Environment

We use the following development environment for this project:

  • Nvidia RTX 3090 GPU
  • Intel Xeon W-1370
  • Ubuntu 20.04
  • CUDA Version 11.2
  • cudatoolkit 11.0
  • Pytorch 1.7
  • Pytorch3D 0.5 or 0.6
  • Detectron2 0.3

Installation

This code is developed using anaconda3 with Python 3.8 (download here), therefore we recommend a similar setup.

You can simply run the following code in the command line to create the development environment:

$ source setup.sh

For visualizing some demo results or using the data preprocessing code, you need our custom rasterizer. In case the provided x86-64 linux shared object does not work for you, you may install the rasterizer here.

Running the Demo

We provide four sample input images in network/assets folder. The images are captured with a smartphone and then preprocessed to be compatible with ROCA format. To run the demo, you first need to download data and config from this Google Drive folder. Models folder contains the pre-trained model and used config, while Data folder contains images and dataset.

Assuming contents of the Models directory are in $MODEL_DIR and contents of the Data directory are in $DATA_DIR, you can run:

$ cd network
$ python demo.py --model_path $MODEL_DIR/model_best.pth --data_dir $DATA_DIR/Dataset --config_path $MODEL_DIR/config.yaml

You will see image overlay and CAD visualization are displayed one by one. Open3D mesh visualization is an interactive window where you can see geometries from different viewpoints. Close the Open3D window to continue to the next visualization. You will see similar results to the image above.

For headless visualization, you can specify an output directory where resulting images and meshes are placed:

$ python demo.py --model_path $MODEL_DIR/model_best.pth --data_dir $DATA_DIR/Dataset --config_path $MODEL_DIR/config.yaml --output_dir $OUTPUT_DIR

You may use the --wild option to visualize results with "wild retrieval". Note that we omit the table category in this case due to large size diversity.

Preparing Data

Downloading Processed Data (Recommended)

We provide preprocessed images and labels in this Google Drive folder. Download and extract all folders to a desired location before running the training and evaluation code.

Rendering Data

Alternatively, you can render data yourself. Our data preparation code lives in the renderer folder.

Our project depends on ShapeNet (Chang et al., '15), ScanNet (Dai et al. '16), and Scan2CAD (Avetisyan et al. '18) datasets. For ScanNet, we use ScanNet25k images which are provided as a zip file via the ScanNet download script.

Once you get the data, check renderer/env.sh file for the locations of different datasets. The meanings of environment variables are described as inline comments in env.sh.

After editing renderer/env.sh, run the data generation script:

$ cd renderer
$ sh run.sh

Please check run.sh to see how individual scripts are running for data preprocessing and feel free to customize the data pipeline!

Training and Evaluating Models

Our training code lives in the network directory. Navigate to the network/env.sh and edit the environment variables. Make sure data directories are consistent with the ones locations downloaded and extracted folders. If you manually prepared data, make sure locations in /network/env.sh are consistent with the variables set in renderer/env.sh.

After you are done with network/env.sh, run the run.sh script to train a new model or evaluate an existing model based on the environment variables you set in env.sh:

$ cd network
$ sh run.sh

Replicating Experiments from the Main Paper

Based on the configurations in network/env.sh, you can run different ablations from the paper. The default config will run the (final) experiment. You can do the following edits cumulatively for different experiments:

  1. For P+E+W+R, set RETRIEVAL_MODE=resnet_resnet+image
  2. For P+E+W, set RETRIEVAL_MODE=nearest
  3. For P+E, set NOC_WEIGHTS=0
  4. For P, set E2E=0

Resources

To get the datasets and gain further insight regarding our implementation, we refer to the following datasets and open-source codebases:

Datasets and Metadata

Libraries

Projects

This can be use to convert text in a file to handwritten text.

TextToHandwriting This can be used to convert text to handwriting. Clone this project or download the code. Run TextToImage.py give the filename of th

Ashutosh Mahapatra 2 Feb 06, 2022
EQFace: An implementation of EQFace: A Simple Explicit Quality Network for Face Recognition

EQFace: A Simple Explicit Quality Network for Face Recognition The first face recognition network that generates explicit face quality online.

DeepCam Shenzhen 141 Dec 31, 2022
An Optical Character Recognition system using Pytesseract/Extracting data from Blood Pressure Reports.

Optical_Character_Recognition An Optical Character Recognition system using Pytesseract/Extracting data from Blood Pressure Reports. As an IOT/Compute

Ramsis Hammadi 1 Feb 12, 2022
GDB python tool to pretty print and debug c++ xtensor containers

gdb_xt2np GDB python tool to pretty print, examine, and debug c++ Xtensor containers. Xtensor is a c++ library for scientific computing using multidim

Christopher Burke 4 Oct 29, 2021
Implementation of EAST scene text detector in Keras

EAST: An Efficient and Accurate Scene Text Detector This is a Keras implementation of EAST based on a Tensorflow implementation made by argman. The or

Jan Zdenek 208 Nov 15, 2022
Deskew is a command line tool for deskewing scanned text documents. It uses Hough transform to detect "text lines" in the image. As an output, you get an image rotated so that the lines are horizontal.

Deskew by Marek Mauder https://galfar.vevb.net/deskew https://github.com/galfar/deskew v1.30 2019-06-07 Overview Deskew is a command line tool for des

Marek Mauder 127 Dec 03, 2022
A small C++ implementation of LSTM networks, focused on OCR.

clstm CLSTM is an implementation of the LSTM recurrent neural network model in C++, using the Eigen library for numerical computations. Status and sco

Tom 794 Dec 30, 2022
A webcam-based 3x3x3 rubik's cube solver written in Python 3 and OpenCV.

Qbr Qbr, pronounced as Cuber, is a webcam-based 3x3x3 rubik's cube solver written in Python 3 and OpenCV. 🌈 Accurate color detection 🔍 Accurate 3x3x

Kim 金可明 502 Dec 29, 2022
An application of high resolution GANs to dewarp images of perturbed documents

Docuwarp This project is focused on dewarping document images through the usage of pix2pixHD, a GAN that is useful for general image to image translat

Thomas Huang 97 Dec 25, 2022
Textboxes implementation with Tensorflow (python)

tb_tensorflow A python implementation of TextBoxes Dependencies TensorFlow r1.0 OpenCV2 Code from Chaoyue Wang 03/09/2017 Update: 1.Debugging optimize

Jayne Shin (신재인) 20 May 31, 2019
PSENet - Shape Robust Text Detection with Progressive Scale Expansion Network.

News Python3 implementations of PSENet [1], PAN [2] and PAN++ [3] are released at https://github.com/whai362/pan_pp.pytorch. [1] W. Wang, E. Xie, X. L

1.1k Dec 24, 2022
Automatically remove the mosaics in images and videos, or add mosaics to them.

Automatically remove the mosaics in images and videos, or add mosaics to them.

Hypo 1.4k Dec 30, 2022
Repositório para registro de estudo da biblioteca opencv (Python)

OpenCV (Python) Objetivo do Repositório: Registrar avanços no estudo da biblioteca opencv. O repositório estará aberto a qualquer pessoa e há tambem u

1 Jun 14, 2022
OCR-D-compliant page segmentation

ocrd_segment This repository aims to provide a number of OCR-D-compliant processors for layout analysis and evaluation. Installation In your virtual e

OCR-D 59 Sep 10, 2022
Image Smoothing and Blurring Using OpenCV

Image-Smoothing-and-Blurring-Using-OpenCV This repository contains codes for performing image smoothing and blurring using OpenCV. There are different

Happy N. Monday 3 Feb 15, 2022
👄 The most accurate natural language detection library for Java and the JVM, suitable for long and short text alike

Quick Info this library tries to solve language detection of very short words and phrases, even shorter than tweets makes use of both statistical and

Peter M. Stahl 532 Dec 28, 2022
Rubik's Cube in pygame with OpenGL

Rubik Rubik's Cube in pygame with OpenGL The script show on the screen a Rubik Cube buit with OpenGL. Then I have also implemented all the possible mo

Gabro 2 Apr 15, 2022
An OCR evaluation tool

dinglehopper dinglehopper is an OCR evaluation tool and reads ALTO, PAGE and text files. It compares a ground truth (GT) document page with a OCR resu

QURATOR-SPK 40 Dec 20, 2022
Convolutional Recurrent Neural Network (CRNN) for image-based sequence recognition.

Convolutional Recurrent Neural Network This software implements the Convolutional Recurrent Neural Network (CRNN), a combination of CNN, RNN and CTC l

Baoguang Shi 2k Dec 31, 2022
Recognizing cropped text in natural images.

ASTER: Attentional Scene Text Recognizer with Flexible Rectification ASTER is an accurate scene text recognizer with flexible rectification mechanism.

Baoguang Shi 681 Jan 02, 2023