Posted on

Jana’s Status Report for 03/08/2025

This week, I worked on my part of the design report which included the abstract, introduction, use-case requirements, design requirements, ML-specific design trade studies, ML-specific testing and validation, schedule and task division, and the related works section. 

On the technical side, I completed labeling the datasets consistently, ensuring uniformity across images from multiple sources. I also finalized the data augmentation techniques, including rotation, greyscale conversion, and flipping, and have begun implementing them. As a result, I now have a fully labeled dataset ready for model training.

Additionally, I established a plan for testing live plants. I am monitoring eight plants under different conditions to support model training and evaluation. This includes 2 plants being underwatered, 2 being overwatered, 2 with nutrient deficiencies, and 2 healthy plants.

I am slightly behind schedule as I have not yet established the initial training framework for the model due to midterm exams and other deadlines. To catch up, I will continue to work through spring break and dedicate additional hours over the next week to make progress.

Next Week’s Deliverables:

  • Establish the initial training framework for the model
  • Test different model architectures (ResNet18 vs MobileNet) using the available dataset
  • Compare model performance on greyscale vs RGB images
  • Continue monitoring live plants for testing and validation
Posted on

Jana’s Status Report for 02/22/2025

This week, I worked on my part of the design presentation which included defining the quantitative design requirements. I also chose and ordered the live plants for the project, which means we now have a clear direction for data collection and analysis.

For the plant health prediction ML model, I selected the datasets to be used in training, opting for images of entire plants instead of just leaves as originally planned. As many of the available datasets didn’t have all the necessary data, I chose to combine multiple datasets to include the required abiotic and biotic stress factors. Since the dataset consists of full plant images, the leaf detection step may not be necessary for now. Instead, I plan to train the ML model on these images directly and will only preprocess the images if needed, based on model performance.

I explored several potential models for plant detection and health classification. YOLOv8n/YOLOv8s stand out as suitable options for object detection. These models would allow the identification and labeling of multiple plants and their individual parts within the images, which is suitable for our 3 plant system. Additionally, I’m considering ResNet50 for better overall performance or MobileNet, which is more lightweight and better suited for deployment on an RPi 5. My plan is to train multiple models and evaluate which one performs the best.

I am on schedule.

Next Week’s Deliverables:

  • I will start labeling the datasets consistently, given that the images come from multiple sources. This will ensure that the training data is standardized. 
  • I will also perform data augmentation to create a larger training set.
  • I will also establish the initial training framework for the model.
Posted on

Team Status Report for 02/15/2025

The most significant risk this week was potential plant loss during model training or system adjustments, which could disrupt data collection. We are managing this by scaling up the system to support three plants instead of one, ensuring a larger dataset and providing backup in case any plants die. Additionally, we are mitigating this risk by setting up the sensors as quickly as possible to begin data collection early. This was the most significant change to the design.

Challenges & Mitigation:

  • Conflicting sensor data: To ensure uniform readings, we placed all plants in a single large pot, allowing us to use one set of sensors and actuators for all plants.
  • Limited Raspberry Pi pins: We selected multi-function sensors (e.g., a 3-in-1 soil pH, nutrient, moisture sensor, and a 2-in-1 temperature/humidity sensor) to reduce the number of connections needed.
  • Increased costs: A larger greenhouse ($103.99) and increased water capacity were necessary to support three plants. To offset this, we are minimizing additional costs by reducing the number of sensors and opting for cost-effective components.

Progress:

  • Finalized and ordered all system components.
  • Worked on the design presentation. 
  • Began planning the software, including machine learning integration and the web application.

Next steps:

  • Set up the Raspberry Pi 5, sensors, and camera once the components arrive.
  • Finalize the data collection pipeline for ML training.
  • Develop the web app framework and user interface.

We will include photos of the system setup once the components arrive.

 

Part A: Our automated greenhouse system enhances public health, safety, and welfare by providing an efficient, low-maintenance way to grow fresh produce, improving food security and reducing reliance on commercially farmed crops with chemical treatments. Safety is a top priority, and we are implementing physical dividers to separate water-sensitive electrical components from wet areas like the water tank, minimizing the risk of short circuits or electrical hazards. Additionally, the system automates irrigation and nutrient delivery, preventing overwatering, mold growth, and manual handling risks. With live monitoring and an intuitive control interface, the system reduces the need for constant supervision, making it safer and more accessible for users with limited mobility or time constraints. By offering an affordable and sustainable way to grow plants, our solution supports self-sufficiency and environmental sustainability while ensuring a safer, healthier growing environment.

Part B: In terms of social factors, our project is significant in that it promotes sustainable urban living by allowing easy indoor plant care. People who cannot be physically present at home to take care of the plants can take advantage of our solution. It is also especially impactful in communities that have limited access to green spaces, as people can grow their plants and form their personal gardens inside their houses. People that share the same sustainability goals regarding plant care can connect more, and educate each other with the plant information that our solution provides. Sproutly contributes to achieving more green-friendly and socially connected society.

Part C: Our goal is to create a mid-range product that balances affordability and advanced features. High-end competitors like AC Infinity ($699 for a 3 plant system) and Plant Hive ($984 for 1 plant) offer advanced sensing and automation but at a steep price. Our system provides similar capabilities at a more accessible cost of $385. This is comparable to Koru ($400), a lower-cost option that offers basic sensors and watering schedules but lacks control over external conditions. Unlike cheaper alternatives that support only one plant in an open environment, our system accommodates three or more plants in an enclosed space with real-time sensing, monitoring, camera feed, and ML-based plant health analysis. This makes our system a cost-effective alternative with premium features at a lower price point.

Part A was written by Zara, B was written by Yuna, and C was written by Jana.

 

Posted on

Jana’s Status Report for 02/15/2025

I researched and finalized the components required for the system, including all the sensors and the camera. I ensured that these components met the project’s durability and cost requirements and considered the pin requirements for connection to the Raspberry Pi 5.

I also worked on the design presentation, where I was responsible for updating the project schedule. Additionally, I researched the quantitative design requirements and linked them to the use case requirements wherever applicable.

I began researching potential plant health detection model datasets, including sensor and image datasets. While I found many datasets related to crop diseases and general health classification, I encountered a lack of suitable ones focused on household plants, specifically addressing conditions like over/under-watering, nutrient deficiencies, or excessive heat. Most datasets are centered on crop diseases or pest detection, which may not fully align with our needs.

I researched models that could be used for training the machine learning model, including pre-trained CNN models such as ResNet and MobileNet, as well as MLP models. I identified possible relevant works.

My progress is on schedule. I have completed the component research task and I am on track with my responsibilities for dataset research.

Next Week’s Deliverables:

  • Finalize dataset and model selection. If necessary, I will explore the option of training from my own dataset, though I am unsure if that is feasible.
  • Begin working on leaf detection using OpenCV.
  • Work on my assigned sections for the design report.
Posted on

Jana’s Status Report for 02/08/2025

This week, I worked on the proposal presentation, including use cases, use case requirements, testing, task division, and schedule (Gantt chart). Since I was responsible for presenting, I dedicated additional time to practicing and refining the presentation.

I researched existing products to build a well-informed feature set, identifying their strengths and limitations. I conducted informal surveys with friends, both plant owners and non-owners, to understand what features they would find most useful. Based on this feedback, I determined key use case requirements. Furthermore, I investigated the quantitative specifications these features should meet. For instance, I analyzed how much water capacity is required for a plant to remain self-sufficient for up to two weeks.

Additionally, I worked on defining testing and verification requirements with my teammates. I researched various sensors and methods for measuring their accuracy and precision in real-world conditions. I also spent significant time structuring the project schedule. I broke down tasks and submission deadlines into manageable phases, ensuring sufficient integration time and allocating buffer periods for unexpected events such as spring break and potential unforeseen challenges before the final demo.

I also began exploring suitable datasets for the machine learning application. I searched for publicly available plant sensor data and identified a Kaggle dataset that aligns well with our objectives. This dataset includes sensor data that we plan on measuring, along with ground truth labels for plant health status. Moreover, I looked into image datasets for plant health detection and found that most of them focus on leaf analysis. This insight led me to conclude that we will need to preprocess plant images using OpenCV to detect leaves before passing them into the model.

We are currently on schedule.

Next Week’s Deliverables:

  • Finalize the sensors to make the purchases.
  • Finalize the sensor and image datasets for the machine learning model.
  • Explore different machine learning models suitable for plant health detection.
  • Begin working on my assigned sections of the design presentation slides.