Building an Object Detection toolkit with TensorFlow (PyLadies Meetup)

In recent years, models based on Convolutional Neural Networks (CNNs) have revolutionized the entire field of computer vision. Problems like image classification can now be considered solved, and it is easy to construct implementations with any modern Deep Learning framework using fine tuning with pre-trained weights on datasets such as ImageNet. However, the harder problems of object detection and segmentation require much more complex methods to solve. Object detection consists of picking up the objects and drawing a rectangular bounding box, while segmentation aims to identify the exact pixels that belong to each object. One of the main differences with image classification is that the same image may contain several objects and those could be in very different proportions, sizes, lighting and partially occluded. In this talk, we will discuss how state of the art object detection techniques work. In particular, we will take a look at Region Proposal Networks (RPNs), which propose candidate object locations (“proposals”) which are later refined to achieve precise localization. We will then look at the architecture of an object detection system, and the performance considerations of different algorithms. Moreover, we will explore an implementation of an open source Python object detection toolkit based on TensorFlow, going through the details and tricks of the trade when using such ecosystem on production ready applications.

Transcript

1. Building an Object Detection toolkit with TensorFlow From academic papers to open source implementation
2. | @tryolabs Who we are Alan Descoins CTO @dekked_ 2 Introduction
3. Then vs now
4. 4 Introduction Felzenszwalb et al., “Object Detection with Discriminatively Trained Part Based Models”, IEEE Transactions on Pattern Analysis and Machine Intelligence, 32, 2010. Detected objects in a sample image (from the COCO dataset) (2017). Source: Google Research Blog. sofa bottle sofa
5. Agenda 5 Introduction 5 5 Introduction Challenges and applications of object detection Demystifying it: dive into Faster R-CNN Luminoth: our open-source toolkit for computer vision
6. Challenges of object detection 6 Introduction
7. Applications of object detection 7 Introduction CT scan of a lung cancer patient at the Jingdong Zhongmei private hospital in Yanjiao, China's Hebei Province (AP Photo/Andy Wong) Hsieh et al., “Drone-based Object Counting by Spatially Regularized Regional Proposal Networks”, ICCV 2017. Source: Pinterest
8. A hard problem with lots of applications 8 Make it accessible! Build a toolkit! Introduction
9. Deep Learning & Object detection
10. Figure from https://blog.heuritech.com/2016/02/29/a-brief-report-of-the-heuritech-deep-learning-meetup-5/ Convolutional feature map Power of ConvNets as feature extractors Pre-train: 10 Deep Learning & Object detection
11. Regression based methods Type of object detection models 11 Deep Learning & Object detection Region proposal based methods Single stage prediction of object classes and bounding boxes. Examples: • You Only Look Once (YOLO) • Single Shot MultiBox Detector (SSD) Two stages: 1. Generate candidate locations using some algorithm. 2. Adjustment of bounding boxes and classification. Examples: • R-CNN, Fast R-CNN, Faster R-CNN
12. Faster R-CNN
13. Background Evolution of methods proposed in previous years: 13 Faster R-CNN 2014 R-CNN - Girshick et al. 2015 Fast R-CNN - Girshick. 2016 Faster R-CNN Ren, Girshick et al. “Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks”, CVPR 2016.
14. 14 Faster R-CNN
15. 15 Overview Faster R-CNN RoIP 3. Region of Interest (RoI) Pooling R-CNN 4. Region-based CNN (R-CNN) RPN 2. Region Proposal Network (RPN) 1. Pre-trained base network
16. Faster R-CNN 1. Pre-trained base network
17. Pre-trained base network Image of arbitrary size → feature map. Common architectures: • VGG (16, 19) • ResNet (v1, v2) • Inception (V2, V3) Feature map encodes information for object detection. 17 Faster R-CNN Feature map 50 37 600 800 CNN (VGG16) 3 512
18. Faster R-CNN 2. Region Proposal Network (RPN)
19. Region proposals 19 Image (feature map) → proposals: • variable number. • different scales and aspect ratios. • efficient process • project bounding boxes to original image. Idea: start with reference boxes, later adjust. How many reference boxes? A lot! Faster R-CNN
20. Anchor boxes For each spatial position of the feature map, generate k fixed anchors (with same center). 20 Faster R-CNN 3 scales, 3 aspect ratios (k=9)
21. Anchor centers in original image Anchors reference (9 anchors per position) Visualizing anchor boxes (1) Anchors on top of single point 21 Faster R-CNN
22. All anchors superimposed Visualizing anchor boxes (2) Ground truth boxes labels: person, bicycle 22 Faster R-CNN
23. Region Proposal Network (RPN) Feature map → rectangular proposals + “objectness” score 23 Faster R-CNN RPN 3x3 conv (pad 1, 512 output channels) 1x1 conv (2k output channels) 1x1 conv (4k output channels) 2k objectness scores 4k box regression scores
24. All positive anchors IoU > 0.7 Anchors batch positive (green), negative (red) 24 Faster R-CNN RPN anchor targets Need positive (foreground) vs negative (background) anchors. Use Intersection over Union (IoU) with ground truth. Faster R-CNN
25. What’s missing 25 Faster R-CNN Multi-task loss Filtering of proposals • Use Non-Maximum Suppression (NMS). • Keep top in “objectness” only. Classification Standard logarithmic loss for 2 classes. Box regression Smooth L1 between difference of coordinates (positive anchors).
26. 26 Faster R-CNN
27. Faster R-CNN 3. Region of Interest (RoI) Pooling
28. RoI Pooling layer 28 Faster R-CNN Arbitrarily-sized proposals → fixed spatial size • Can feed output to fully connected layers. • Very similar to max pooling. Faster R-CNN Project RoI Pool 7x7x512 Proposal RoI 512
29. Faster R-CNN 4. Region-based CNN (R-CNN)
30. Region-based CNN (R-CNN) 30 Faster R-CNN Fixed-size outputs of RoI Pooling→ Faster R-CNN 7x7x512 probability distribution (N+1 classes) bounding box regressions (N classes) Flatten FC FC bicycle p=0.96 Softmax
31. 31 Faster R-CNN person (0.99) bicycle (0.97)
32. Building a toolkit
33. Building a toolkit What is Luminoth? Open-source deep learning library/toolkit for computer vision object detection. 33 CLI tools Pre-defined models Cloud integration
34. $ pip install luminoth $ lumi train # Magic The goal 34 Building a toolkit
35. Building a toolkit Objectives 35 “Out-of-the-box” usage Production ready Open source Readable code Extensible and modular
36. Building a toolkit TensorFlow + Sonnet import sonnet as snt def RPN(snt.AbstractModule): def __init__(self, *args, name='rpn'): [...] # submodules init, config def _build(self, inputs): # TensorFlow code. return outputs 36 +
37. Building a toolkit “Model oriented programming” • Follow OOP good practices Faster R-CNN RPN in: → feature map, anchors out: → proposals RCNN in: → proposals, pooled feature maps out: → objects, labels, probabilities 37
38. ObjectDetection 38 Hierarchical structure Building a toolkit RPN R-CNN RoIP FasterRCNN RPN RPNTargets RPNProposals TFRecordDataset ObjectDetectionDataset RoIPooling RCNN RCNNTargets RCNNProposals TruncatedNetwork VGG/ResNet
39. Building a toolkit Challenges of coding from papers 39 Small implementation details have no room in academic papers Papers tend to remain frozen in time Many ways to implement it
40. Building a toolkit Challenges of Faster R-CNN implementation 40 Multiple moving parts Module dependencies Multi-task training
41. Data pipeline Debugging Training Data visualization Evaluation Deployment Beyond the model Distributed 41 Building a toolkit Unit testing Monitoring Model
42. Building a toolkit https://github.com/tryolabs/luminoth Using Luminoth 42 $ pip install luminoth $ lumi --help Usage: lumi [OPTIONS] COMMAND [ARGS]... Options: -h, --help Show this message and exit. Commands: cloud Groups of commands to train models in the cloud dataset Groups of commands to manage datasets evaluate Evaluate trained (or training) models server Groups of commands to serve models train Train models
43. $ lumi dataset transform --type pascal --data-dir /data/pascal --output /data/ # Create tfrecords for optimizing data consumption. $ lumi train --config pascal-fasterrcnn.yml # Hours of training $ tensorboard --logdir jobs/ # On another GPU/Machine/CPU $ lumi evaluate --config pascal-fasterrcnn.yml # Checks for new checkpoints and writes logs # Finally $ lumi server web --config pascal-fasterrcnn.yml # Looks for checkpoint and loads it into a simple frontend/json API server. Luminoth cycle 43 Building a toolkit
44. Building a toolkit Luminoth’s future 44 Fine-tune trained models More models & problems Tagging ↔ Training integration Distributed deployment
45. Thanks for listening! Questions? Learn more & contribute github.com/tryolabs/luminoth