Object Detection and Classification of satellite Images -- 3
Budget: ₹1,500 – ₹12,500 INR
I have the final requirement listed here
Object Detection of various objects such as cars(different types), bridges, roads (black top, cemented, tracks etc), shoreline, Ports, trucks (Different types), buildings (different types), radio stations (different types), ships (different types), airstrips, planes (different types), vegitation (Forests, scrubs etc), farmlands (different types if possible), barren lands etc from satellite images. Should be compatible with QGIS.
OS:
Operating system is Ubuntu (Preferred)/Windows 10
Inputs:
1. For the projects Datapoints on google earth in the form of kml file can be shared if required.
2. Data freely available can be utilised.
3. DSTL - Kaggle (https://www.kaggle.com/c/dstl-satellite-imagery-feature-detection)
4. Draper Satellite Image Chronology - Kaggle (https://www.kaggle.com/c/draper-satellite-image-chronology)
5. Understanding the Amazon from Space - Kaggle (https://www.kaggle.com/c/planet-understanding-the-amazon-from-space)
6. Ships in Satellite Imagery - Kaggle - From Open California dataset. (https://www.kaggle.com/rhammell/ships-in-satellite-imagery)
7. 2D Semantic Labeling - Vaihingen data - This is from a true orthophotographic survey (from a plane). The data is 9cm resolution! (http://www2.isprs.org/commissions/comm3/wg4/2d-sem-label-vaihingen.html)
8. SAT-4 and SAT-6 airborne datasets - Very high res (1m) whole of US. 65 Terabytes. (DeepSat) 1m/px. (2015). SAT-4: barren land, trees, grassland, other. SAT-6: barren land, trees, grassland, roads, buildings, water bodies. (RGB + NIR)
(http://csc.lsu.edu/~saikat/deepsat/)
9. The Rareplanes Dataset (https://www.cosmiqworks.org/RarePlanes/)
10. Any other dataset deemed fit and available in the open-source or your own.
Language of programming: Python, Jupyter Notebooks
Pre-Trained Models:
Any open-source pre-trained model may be used to avoid starting from scratch.
Language of programming: Python and/or Jupyter Notebooks. Complete installable files on Ubuntu platform 20.0 or later are preferred. Can be on MS Win-10 too.
Steps of work:
A complete pipeline of project is required. From data collection stage, Training and curation in a big data framework, etc.
Steps of work:
Notebook 1 Creating Dataset (Chips) using JSON/KML (only for collecting chips from google maps if required) File as input or opensource automated using QGIS Plugins.
Notebook 2 Dataset enhancement using GANs and augmentation of the data.
Notebook 3 Training of Deep Neural Network on Dataset with the option to switch from Yolov3/Yolov4/Yolov5/Xview or anyother family or custom models etc. Validation and testing with the ability to change different models.
Notebook 4 Input pipeline to detect objects from given images in a directory. Output should be a bounding boxed image. With a JSON file of objects detected and a .csv file. The user should have the option to check which objects he is interested in detection an independent web application or web-based QGIS application or QGIS
A text File describing all steps neatly for a demonstration on a standalone PC is required to be given.
Notebook 5 Creation of a text report of objects detected using a .csv file.
The API-based web applications should not be on bootstraps or web dependant. The application should store all the processed images into NoSQL along with detected objects and segmentation areas.
The user should be able to search using his image file name or date of processing of the image.
Any paper or open-source work including data used should be referenced in detail. Any project adjustments may be discussed before implementation. Fast and efficient implementation of open-source code is expected.
Budget:
Budgetary constraints exists and hence negotiable.
Object Detection of various objects such as cars(different types), bridges, roads (black top, cemented, tracks etc), shoreline, Ports, trucks (Different types), buildings (different types), radio stations (different types), ships (different types), airstrips, planes (different types), vegitation (Forests, scrubs etc), farmlands (different types if possible), barren lands etc from satellite images. Should be compatible with QGIS.
OS:
Operating system is Ubuntu (Preferred)/Windows 10
Inputs:
1. For the projects Datapoints on google earth in the form of kml file can be shared if required.
2. Data freely available can be utilised.
3. DSTL - Kaggle (https://www.kaggle.com/c/dstl-satellite-imagery-feature-detection)
4. Draper Satellite Image Chronology - Kaggle (https://www.kaggle.com/c/draper-satellite-image-chronology)
5. Understanding the Amazon from Space - Kaggle (https://www.kaggle.com/c/planet-understanding-the-amazon-from-space)
6. Ships in Satellite Imagery - Kaggle - From Open California dataset. (https://www.kaggle.com/rhammell/ships-in-satellite-imagery)
7. 2D Semantic Labeling - Vaihingen data - This is from a true orthophotographic survey (from a plane). The data is 9cm resolution! (http://www2.isprs.org/commissions/comm3/wg4/2d-sem-label-vaihingen.html)
8. SAT-4 and SAT-6 airborne datasets - Very high res (1m) whole of US. 65 Terabytes. (DeepSat) 1m/px. (2015). SAT-4: barren land, trees, grassland, other. SAT-6: barren land, trees, grassland, roads, buildings, water bodies. (RGB + NIR)
(http://csc.lsu.edu/~saikat/deepsat/)
9. The Rareplanes Dataset (https://www.cosmiqworks.org/RarePlanes/)
10. Any other dataset deemed fit and available in the open-source or your own.
Language of programming: Python, Jupyter Notebooks
Pre-Trained Models:
Any open-source pre-trained model may be used to avoid starting from scratch.
Language of programming: Python and/or Jupyter Notebooks. Complete installable files on Ubuntu platform 20.0 or later are preferred. Can be on MS Win-10 too.
Steps of work:
A complete pipeline of project is required. From data collection stage, Training and curation in a big data framework, etc.
Steps of work:
Notebook 1 Creating Dataset (Chips) using JSON/KML (only for collecting chips from google maps if required) File as input or opensource automated using QGIS Plugins.
Notebook 2 Dataset enhancement using GANs and augmentation of the data.
Notebook 3 Training of Deep Neural Network on Dataset with the option to switch from Yolov3/Yolov4/Yolov5/Xview or anyother family or custom models etc. Validation and testing with the ability to change different models.
Notebook 4 Input pipeline to detect objects from given images in a directory. Output should be a bounding boxed image. With a JSON file of objects detected and a .csv file. The user should have the option to check which objects he is interested in detection an independent web application or web-based QGIS application or QGIS
A text File describing all steps neatly for a demonstration on a standalone PC is required to be given.
Notebook 5 Creation of a text report of objects detected using a .csv file.
The API-based web applications should not be on bootstraps or web dependant. The application should store all the processed images into NoSQL along with detected objects and segmentation areas.
The user should be able to search using his image file name or date of processing of the image.
Any paper or open-source work including data used should be referenced in detail. Any project adjustments may be discussed before implementation. Fast and efficient implementation of open-source code is expected.
Budget:
Budgetary constraints exists and hence negotiable.