Image Object Detection Model for Random Sample Images Using TensorFlow Take 6

Template Credit: Adapted from an Object Detection tutorial on TensorFlow.org.

Additional Notes: I adapted this workflow from the TensorFlow Object Detection tutorial on TensorFlow.org. I plan to build a script for building future projects using object detection models.

SUMMARY: This project aims to construct an object detection model using the TensorFlow-based neural network and document the end-to-end steps using a template.

This iteration will use the TF2 Mask R-CNN Inception ResNet V2 1024×1024 object detection model to test some sample images. The model was constructed using the Mask R-CNN Object detection model and trained on COCO 2017 dataset with training images scaled to 1024×1024.

Images Used: 1. Airport runway; 2. Tanzania Safari; 3. Streets with Cars; 4. Public Library

Dataset ML Model: Image Object Detection using TensorFlow Hub Models

Additional References: https://tfhub.dev/s?module-type=image-object-detection

The HTML formatted report can be found here on GitHub.

Multi-Class Model for Crop Mapping with Fused Optical and Radar Data Using TensorFlow

Template Credit: Adapted from a template made available by Dr. Jason Brownlee of Machine Learning Mastery.

SUMMARY: This project aims to construct a predictive model using various machine learning algorithms and document the end-to-end steps using a template. The Crop Mapping with Fused Optical Radar Data dataset is a multi-class modeling situation where we attempt to predict one of several (more than two) possible outcomes.

INTRODUCTION: This dataset combines optical and PolSAR remote sensing images for cropland classification. The organization collected the images using RapidEye satellites (optical) and the Unmanned Aerial Vehicle Synthetic Aperture Radar (UAVSAR) system (radar) over an agricultural region near Winnipeg, Manitoba, Canada, in 2012. There are two sets of 49-radar features and two sets of 38-optical features for 05 and 14 July 2012. Seven crop type classes exist for this data set: 1-Corn; 2-Peas; 3-Canola; 4-Soybeans; 5-Oats; 6-Wheat; and 7-Broadleaf.

ANALYSIS: The performance of the preliminary TensorFlow models achieved an average accuracy benchmark of 0.9942 after running for 20 epochs. When we applied the final model to the test dataset, the model achieved an accuracy score of 0.9951.

CONCLUSION: In this iteration, the simple TensorFlow model appeared to be a suitable algorithm for modeling this dataset.

Dataset Used: Crop Mapping with Fused Optical Radar Data

Dataset ML Model: Multi-class classification with numerical attributes

Dataset Reference: https://archive-beta.ics.uci.edu/ml/datasets/crop+mapping+using+fused+optical+radar+data+set

The HTML formatted report can be found here on GitHub.

Multi-Class Model for Crop Mapping with Fused Optical and Radar Data Using TensorFlow Decision Forests

Template Credit: Adapted from a template made available by Dr. Jason Brownlee of Machine Learning Mastery.

SUMMARY: This project aims to construct a predictive model using various machine learning algorithms and document the end-to-end steps using a template. The Crop Mapping with Fused Optical Radar Data dataset is a multi-class modeling situation where we attempt to predict one of several (more than two) possible outcomes.

INTRODUCTION: This dataset combines optical and PolSAR remote sensing images for cropland classification. The organization collected the images using RapidEye satellites (optical) and the Unmanned Aerial Vehicle Synthetic Aperture Radar (UAVSAR) system (radar) over an agricultural region near Winnipeg, Manitoba, Canada, in 2012. There are two sets of 49-radar features and two sets of 38-optical features for 05 and 14 July 2012. Seven crop type classes exist for this data set: 1-Corn; 2-Peas; 3-Canola; 4-Soybeans; 5-Oats; 6-Wheat; and 7-Broadleaf.

ANALYSIS: The performance of the preliminary Gradient Boosted Trees model achieved an accuracy benchmark of 0.9976 on the validation dataset. The final model processed the validation dataset with a final accuracy score of 0.9999. When we applied the finalized model to Kaggle’s test dataset, the model achieved an accuracy score of 0.9996.

CONCLUSION: In this iteration, the TensorFlow Decision Forests model appeared to be a suitable algorithm for modeling this dataset.

Dataset Used: Crop Mapping with Fused Optical Radar Data

Dataset ML Model: Multi-class classification with numerical attributes

Dataset Reference: https://archive-beta.ics.uci.edu/ml/datasets/crop+mapping+using+fused+optical+radar+data+set

The HTML formatted report can be found here on GitHub.

Multi-Class Model for Crop Mapping with Fused Optical and Radar Data Using XGBoost

Template Credit: Adapted from a template made available by Dr. Jason Brownlee of Machine Learning Mastery.

SUMMARY: This project aims to construct a predictive model using various machine learning algorithms and document the end-to-end steps using a template. The Crop Mapping with Fused Optical Radar Data dataset is a multi-class modeling situation where we attempt to predict one of several (more than two) possible outcomes.

INTRODUCTION: This dataset combines optical and PolSAR remote sensing images for cropland classification. The organization collected the images using RapidEye satellites (optical) and the Unmanned Aerial Vehicle Synthetic Aperture Radar (UAVSAR) system (radar) over an agricultural region near Winnipeg, Manitoba, Canada, in 2012. There are two sets of 49-radar features and two sets of 38-optical features for 05 and 14 July 2012. Seven crop type classes exist for this data set: 1-Corn; 2-Peas; 3-Canola; 4-Soybeans; 5-Oats; 6-Wheat; and 7-Broadleaf.

ANALYSIS: The performance of the preliminary XGBoost model achieved an accuracy benchmark of 0.9977. After a series of tuning trials, the final model processed the training dataset with an accuracy score of 0.9982. When we processed the test dataset with the final model, the model achieved an accuracy score of 0.9982.

CONCLUSION: In this iteration, the XGBoost model appeared to be a suitable algorithm for modeling this dataset.

Dataset Used: Crop Mapping with Fused Optical Radar Data

Dataset ML Model: Multi-class classification with numerical attributes

Dataset Reference: https://archive-beta.ics.uci.edu/ml/datasets/crop+mapping+using+fused+optical+radar+data+set

The HTML formatted report can be found here on GitHub.

Multi-Class Model for Crop Mapping with Fused Optical and Radar Data Using Scikit-learn

Template Credit: Adapted from a template made available by Dr. Jason Brownlee of Machine Learning Mastery.

SUMMARY: This project aims to construct a predictive model using various machine learning algorithms and document the end-to-end steps using a template. The Crop Mapping with Fused Optical Radar Data dataset is a multi-class modeling situation where we attempt to predict one of several (more than two) possible outcomes.

INTRODUCTION: This dataset combines optical and PolSAR remote sensing images for cropland classification. The organization collected the images using RapidEye satellites (optical) and the Unmanned Aerial Vehicle Synthetic Aperture Radar (UAVSAR) system (radar) over an agricultural region near Winnipeg, Manitoba, Canada, in 2012. There are two sets of 49-radar features and two sets of 38-optical features for 05 and 14 July 2012. Seven crop type classes exist for this data set: 1-Corn; 2-Peas; 3-Canola; 4-Soybeans; 5-Oats; 6-Wheat; and 7-Broadleaf.

ANALYSIS: The average performance of the machine learning algorithms achieved an accuracy benchmark of 0.9908 using the training dataset. Furthermore, we selected Extra Trees as the final model as it processed the training dataset with a final accuracy score of 0.9975. When we processed the test dataset with the final model, the model achieved an accuracy score of 0.9976.

CONCLUSION: In this iteration, the Extra Trees model appeared to be a suitable algorithm for modeling this dataset.

Dataset Used: Crop Mapping with Fused Optical Radar Data

Dataset ML Model: Multi-class classification with numerical attributes

Dataset Reference: https://archive-beta.ics.uci.edu/ml/datasets/crop+mapping+using+fused+optical+radar+data+set

The HTML formatted report can be found here on GitHub.

Seth Godin on Survival Is Not Enough, Part 5

In his book, Survival Is Not Enough: Why Smart Companies Abandon Worry and Embrace Change, Seth Godin discusses how innovative organizations and individuals can apply prudent strategies in adapting and positioning themselves for the constant changes.

These are some of my favorite concepts and takeaways from reading the book.

Chapter 5, Your Company Has mDNA

In this chapter, Seth discusses a company’s meme DNA (mDNA) and how organizations can leverage the mDNA metaphor to embrace and adopt changes. He offers the following observations and recommendations for us to think about:

  • A company’s mDNA comprises the rules, processes, policies, market position, and people in the company. Without this mDNA, an organization would forget, from day to day, what it did and how it operated yesterday. What an organization did before today often plays a significant role in planning to do business tomorrow.
  • Unlike human DNA, the company mDNA can mutate as often as the business wants it to. The mDNA must change before the organization can change. Trying to change a business and its people without mutating the mDNA is not possible.
  • The business ecosystem is made up of many businesses. Each business in that environment is a smaller system, but all of them are in the same environment, entwined with each other. At each step along the way, a business, like organisms in nature, evolves along with its people and sub-organizations.
  • The job of the CEO is not to be right about the future as it is impossible to get everything bit right. The position of the CEO is to organize the company to execute a strategy that is winning for now, at the same time, to organize the company to evolve often enough to find the next strategy before today’s plan becomes useless.
  • Companies that can zoom are more likely to evolve, more likely to be launching innovative new products and services, and more likely to be successful. In addition, a company that knows how to zoom more than likely will attract employees who want to zoom. “When your company starts hiring zoomers, it’s going to zoom faster!”
  • Most of the fantastic big business ideas did not initially come from start-ups. Instead, they came from the research group inside big companies that were too rigid to do anything with them. “The challenge companies face is not in inventing new ideas. It’s in moving the old ideas out of the way so that they can implement the new ones.”

In summary:

“Organizations can put the proven tactics of evolution to use by embracing change, not fighting it. By incorporating adopting successful new memes into a company’s mDNA, organizations can defeat their slower competitors.”

你感覺得富裕嗎?

(從我一個尊敬的作家,賽斯·高汀

這和有家財萬貫是不一樣的。

富裕總是相對的。與你的曾祖父母相比,我們是他們簡直不可想像超自然的富有。我們可以獲得一個世紀前無法想像的信息和技術。同時,與十英里外或十年後的人相比,我們已經落後了。

擁有完全相同資源和選擇的兩個人,可能會以完全不同的方式來回答這“富裕”的問題。因為錢終究只是一個故事。

您選擇的社區或行業或同齡人群體,與您是否相對來講富裕有很大關係。

股市調整後,億萬富翁對慈善事業的捐贈減少了。他們的錢仍然多得數不過來,但他們不像以前那麼富有,而且不是那麼富有很容易就被解釋成為不富有。

這意味著對於很多人來說,感覺富裕是一種選擇。

如果這種選擇鼓勵我們去變得專橫、自私和欺負人,那麼最好避免它。

另一方面,如果去選擇將我們的選擇、機會和特權視為通往慷慨、長期思考和聯繫的道路,那麼我們現在就可以做到。

Multi-Class Model for Kaggle Tabular Playground Series 2021 December Using TensorFlow Decision Forests

Template Credit: Adapted from a template made available by Dr. Jason Brownlee of Machine Learning Mastery.

SUMMARY: This project aims to construct a predictive model using various machine learning algorithms and document the end-to-end steps using a template. The Kaggle Tabular Playground December 2021 dataset is a multi-class modeling situation where we attempt to predict one of several (more than two) possible outcomes.

INTRODUCTION: Kaggle wants to provide an approachable environment for relatively new people in their data science journey. Since January 2021, they have hosted playground-style competitions on Kaggle with fun but less complex, tabular datasets. The dataset used for this competition is synthetic but based on a real dataset and generated using a CTGAN. The dataset is used for this competition is synthetic but based on a real dataset and generated using a CTGAN. This dataset is based on the original Forest Cover Type Prediction competition.

ANALYSIS: The performance of the preliminary Gradient Boosted Trees model achieved an accuracy benchmark of 0.9599 on the validation dataset. The final model processed the validation dataset with a final accuracy score of 0.9627. When we applied the finalized model to Kaggle’s test dataset, the model achieved an accuracy score of 0.9535.

CONCLUSION: In this iteration, the TensorFlow Decision Forests model appeared to be a suitable algorithm for modeling this dataset.

Dataset Used: Kaggle Tabular Playground 2021 December Data Set

Dataset ML Model: Multi-Class classification with numerical and categorical attributes

Dataset Reference: https://www.kaggle.com/c/tabular-playground-series-dec-2021

One potential source of performance benchmark: https://www.kaggle.com/c/tabular-playground-series-dec-2021/leaderboard

The HTML formatted report can be found here on GitHub.