# Phase 1 of BLOS Improvement Project - Data Collection and Labeling
###### tags: `log`

[TOC]

## Dataset Collection

Todo:
- [ ] Finish Get Image Code
- [ ] Choose Routes
- [ ] Collect Data
- [ ] Upload Dataset to Roboflow

### Google Street View API
We use a [module from GitHub](https://github.com/hhe1ibeb/streetview) which saved us a lot of time

### Determine the Roads to Collect Data From

考慮條件：
* **無**自行車道的大馬路
台北市自行車道分佈圖

結論：由圖觀察，文山區的自行車道最少

影響自行車安全性的因素
* 斑馬線
* 分隔島 / 停車格

## Training the Model

Todo:
- [ ] Read Papers
- [ ] Compare Models
- [ ] Label Data
- [ ] Train

### Model Comparison
[some ref i haven't read yet](https://medium.com/ching-i/影像分割-image-segmentation-語義分割-semantic-segmentation-1-53a1dde9ed92) about difference between image segmentation and semantic segmentation
[Efficient Deep Models for Monocular Road Segmentation](http://ais.informatik.uni-freiburg.de/publications/papers/oliveira16iros.pdf?ref=https://githubhelp.com)

### Method Comparison

### Paper
* [Ridesharing accessibility from the human eye: Spatial modeling of built environment with street-level images-2022](https://reader.elsevier.com/reader/sd/pii/S0198971522001028?token=E1BEDFEC0AD7A12808FE52CCA157D5F40DD9F2524FF18D1B140B2D48C9B7ED45EC9A5A56370DA6FF41948992C1547577&originRegion=us-east-1&originCreation=20230409131644)
* [Development of a bikeability assessment tool for Dutch cities-2022](https://pure.tue.nl/ws/portalfiles/portal/201242204/Wout_0964113_USRE_Vries_d.pdf)
* [Assessing bikeability with street view imagery and computer vision-2021](https://arxiv.org/pdf/2105.08499.pdf)
Use SVI to extract information of data used in previous bikeability studies under four categories that have been delineated by this study：
* environment 環境
* slope
* urban greenery
* land use
* building design
* connectivity

* infrastructure 基礎設施
* sidewalk
* pavement
* path obstruction
* curb cuts
* crossing
* sidewalk buffer
* street amenity
* bikelane
* road width
[The development and testing of an audit for the pedestrian environment](https://d1wqtxts1xzle7.cloudfront.net/46806200/The_development_and_testing_of_an_audit_20160626-23259-i8vcvj-libre.pdf?1466956134=&response-content-disposition=inline%3B+filename%3DThe_development_and_testing_of_an_audit.pdf&Expires=1681435179&Signature=BamHocO~Lrbk3nabY~W~QKTNJZrWmA3eph2tPbpfkYdq340j3SboGT5JEryNP50TAqlVCiXIfiXKsEqYNEHV5kFmwGJ5OUWzmu8sKiCRAC~vwZS-muckxkiZCXjGLxxAyibt6a7fadB~rhLmFGYlqlkQ79feWZoMHGopY9ma5sdIbTXqw8hWv6F~ptPaTb5KnQL5xb7uhfdf6uAhz8l1OvlkL1j0CbGxq6hDui~08YrcqRAhDNyTESxh6lH2N9kAA0KbGopg7NBuaNAYvnxqeV9p2FMyFvLxaFKjm6x9Mo-ZMuDoPN9yJk0sU897FDPFCN4mmtGmK~45MXCB5w89jw__&Key-Pair-Id=APKAJLOHF5GGSLRBV4ZA)
* vehicle-cyclist interaction ⾞輛與騎⾞⼈互動
* dynamic traffic volume
* [Estimating pedestrian volume using Street View images: A large-scale validation test-2020](https://)
* SVI can be used to explain more than 65 percent of the spatiotemporal mobility pattern
* [Knowledge and topology: A two layer spatially dependent graph neural networks to identify urban functions with time-series street view image-2023](https://www.sciencedirect.com/science/article/abs/pii/S0924271623000680)
* traffic speed
* street parking
* traffic control

* perception 感知
* attractive
* safety
* cleanliness
* crowdedness
* beauty



##### Future directions listed:
1. Incorporate predictive modelling to estimate AQI for each sample point
2. Survey participants control
3. Developing GUI software and API

* [Knowledge and topology: A two layer spatially dependent graph neural networks to identify urban functions with time-series street view image-2023](https://www.sciencedirect.com/science/article/abs/pii/S0924271623000680)
* [Expanding the Scope of the Bicycle Level-of-Service Concept: A Review of the Literature](https://www.researchgate.net/publication/340488257_Expanding_the_Scope_of_the_Bicycle_Level-of-Service_Concept_A_Review_of_the_Literature)
* [Development of Artificial Intelligence-based Bicycle Level of Service Models for Urban Street Segments](https://link.springer.com/article/10.1007/s13177-021-00280-3)
* [Development of a bicycle level of service model for urban street segments in mid-sized cities carrying heterogeneous traffic: A functional networks approach](https://www.sciencedirect.com/science/article/pii/S209575641630294X)
* [Defining Bicycle Levels of Service Criteria Using Levenberg–Marquardt and Self-organizing Map Algorithms](https://www.researchgate.net/publication/327008801_Defining_Bicycle_Levels_of_Service_Criteria_Using_Levenberg-Marquardt_and_Self-organizing_Map_Algorithms)
* [Expanding the Scope of the Bicycle Level-of-Service Concept: A Review of the Literature](https://www.researchgate.net/publication/340488257_Expanding_the_Scope_of_the_Bicycle_Level-of-Service_Concept_A_Review_of_the_Literature)
* [Bicycle Level of Service for Route Choice—A GIS Evaluation of Four Existing Indicators with Empirical Data](https://www.researchgate.net/publication/333000021_Bicycle_Level_of_Service_for_Route_Choice-A_GIS_Evaluation_of_Four_Existing_Indicators_with_Empirical_Data)
#### 評估指數
1. ⾃⾏⾞安全指數評級（BSIR）
2. 道路路段指數（RSI）：考量路段各種參數：交通量、車道數、限速、外側車道寬度、路面因素、位置因素
3. 交叉⼝評價指數（IEI）模型：考慮交叉⼝參數來評估道路交叉⼝的性能
4. Bicycle Suitability Rating (BSR)模型也是基於RSI模型的概念背景開發的
5. 交互⾵險評分（IHS）
* 路邊⼟地利⽤強度
* 路緣切割頻率

#### SAM by Meta
[Segment Anything](https://ai.facebook.com/research/publications/segment-anything/)
[GitHub](https://github.com/facebookresearch/segment-anything)

## Datasets
* [Cityscapes](https://www.cityscapes-dataset.com)
* [Mapillary Vistas](https://www.mapillary.com/dataset/vistas)
* [Place Pulse 2.0](https://paperswithcode.com/dataset/place-pulse-2-0)