Install dependecies:
pip install requirements.txt or pip install requirements_mac.txt for macOS
Download Data from DVC:
dvc pull data - Will download all the data (both video seq and dataset)
if wanted to download only the video sequence execute
dvc pull data/video
Might require log in to the Google Account
In order to upload new data files into the cloud execute commands:
dvc add data/
dvc push
this will update the .dvc file REMEMBER TO PUSH THE NEW .DVC FILE INTO GIT AFTER PUSIING DATA !!!
The data used for the project is a subset of the (COCO dataset)[https://cocodataset.org/#home].
The subset was created using only person, car and bicycle labels.
To generate the dataset you can use src/data/make_dataset.py.
The number of samples , labels and output path can be selected as arguments.
If created a new dataset remember to uploaded to DVC
We used the subset to train a model. The data need to be converted to TFRecord format.
The Sequences are not videos but a set of frames .
- Seq_1: Simple seq with ground truth and no oculusion
- Seq_2: Seq with oclusion and ground truth
- Seq_3: Seq with oclusion no ground truth
Details
In the sequence folder you will find:- Three recorded sequences of pedestrian, cyclists and cars.
- A sequence for camera calibration.
Data format
Each sequence folder contains the following structure:
seq_<n>/
|
|____image02/
| |____data/
| |_ <image_seq_no>.png
| |_ timestamps.txt
|
|
|____image03/
| |____data/
| |_ <image_seq_no>.png
| |_ timestamps.txt
- image_02/data/ contains the left color camera sequence images (png)
- image_02/timestamps.txt timestamps for each left image, the first line being for the first image, the second line for the second, and so on.
- image_03/data/ contains the right color camera sequence images (png)
- image_03/timestamps.txt timestamps for each right image
- labels.txt contains the ground truth labels that can be used to evaluate your solution. Note that we do not provide the labels for sequence 03.
The labels files contain the following information. All values (numerical or strings) are separated via spaces, each row corresponds to one object. The 17 columns represent:
| Values | Name | Description |
|---|---|---|
| 1 | frame |
Frame within the sequence where the object appearers |
| 1 | track id |
Unique tracking id of this object within this sequence |
| 1 | type |
Describes the type of object: Car,Pedestrian, Cyclist
|
| 1 | truncated |
Float from 0 (non-truncated) to 1 (truncated), where truncated refers to the object leaving image boundaries. |
| 1 | occluded |
Integer (0,1,2,3) indicating occlusion state: 0 = fully visible, 1 = partly occluded, 2 = largely occluded, 3 = unknown |
| 1 | alpha |
Observation angle of object, ranging |
| 4 | bbox |
2D bounding box of object in the Rectified image (0-based index): contains left, top, right, bottom pixel coordinates |
| 3 | dimensions |
3D object dimensions: height, width, length (in meters) |
| 3 | location |
3D object location |
| 1 | rotation_y |
Rotation |
| 1 | score |
Only for results: Float, indicating confidence in detection, needed for p/r curves, higher is better. |
Please note that the 2D bounding box provided is given with respect to rectified image coordinates.
- Calibrate and rectify the stereo input.
- Process the input images to detect objects in the environment and track them in 3D, even under occlusions.
- Train a machine learning system that can classify unseen images into the 3 classes (pedestrians, cyclists, and cars) based either on 2D or 3D data. o Use the web or/and capture your own images to create your training set. The image sequences 1 and 2 provided with the project will constitute your validation set and the sequence 3 your testing set Project Organization
File structure
├── LICENSE
├── Makefile <- Makefile with commands like `make data` or `make train`
├── README.md <- The top-level README for developers using this project.
├── data
│ ├── external <- Data from third party sources.
│ ├── interim <- Intermediate data that has been transformed.
│ ├── processed <- The final, canonical data sets for modeling.
│ └── raw <- The original, immutable data dump.
│
├── docs <- A default Sphinx project; see sphinx-doc.org for details
│
├── models <- Trained and serialized models, model predictions, or model summaries
│
├── notebooks <- Jupyter notebooks. Naming convention is a number (for ordering),
│ the creator's initials, and a short `-` delimited description, e.g.
│ `1.0-jqp-initial-data-exploration`.
│
├── references <- Data dictionaries, manuals, and all other explanatory materials.
│
├── reports <- Generated analysis as HTML, PDF, LaTeX, etc.
│ └── figures <- Generated graphics and figures to be used in reporting
│
├── requirements.txt <- The requirements file for reproducing the analysis environment, e.g.
│ generated with `pip freeze > requirements.txt`
│
├── setup.py <- makes project pip installable (pip install -e .) so src can be imported
├── src <- Source code for use in this project.
│ ├── __init__.py <- Makes src a Python module
│ │
│ ├── data <- Scripts to download or generate data
│ │ └── make_dataset.py
│ │
│ ├── features <- Scripts to turn raw data into features for modeling
│ │ └── build_features.py
│ │
│ ├── models <- Scripts to train models and then use trained models to make
│ │ │ predictions
│ │ ├── predict_model.py
│ │ └── train_model.py
│ │
│ └── visualization <- Scripts to create exploratory and results oriented visualizations
│ └── visualize.py
│
└── tox.ini <- tox file with settings for running tox; see tox.readthedocs.io
Project based on the cookiecutter data science project template. #cookiecutterdatascience
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent