Congratulations to Dr. Junhan Wen, for successfully defending her dissertation, titled
"From iMage to Market - Machine-Learning-Empowered Fruit Supply"
Junhan recently won the Open Science award for the Growing Strawberries Dataset that she assembled as part of her PhD.
https://www.tudelft.nl/2025/ewi/st/junhan-wen-receives-tu-delft-open-science-awards-2025
#PhDLife #AcademicLife #AcademicChatter #MachineLearning #OpenScience #TUDelft #PhDone
 
      
  
                            
                        
                        Junhan presents her research on how to predict when an individual strawberry is ripe for harvest, and how to use that information to predict the size of future harvests.
                           
      
  
                            
                        
                        Part of the ceremony. The committee in robes are sitting behind a table, while one member stands up and speeches. Freshly minted doctor and her paranimphs are standing in front of them, with their backs to the viewer, listening. All are dressed in black. TU Delft logo in background, coat of arms on floor.
                           
      
  
                            
                        
                        A pile of printed dissertations. Their cover is an illustration of a robot hand pointing at strawberries, some of them riper than others. Charts illustrate their ripeness.
                           
      
  
                            
                        
                        Propositions
1. An effective multiple object tracking method for associating the observations must dynamically and adaptively combine motion and appearance features. (Chapter 3)
2. Image data alone is insufficient for internal quality estimation, but can augment other information to make predictions for individual fruits. (Chapter 4)
3. The color spectrum serves as a universal criterion for fruit maturity, offering a reliable and transferable method for evaluation by both humans and robots. (Chapter 5)
4. Training a model using the regret of its decisions leads to an appropriate level of conservatism. (Chapter 6)
5. A research project is like a Lego set: only a few bricks are visible, while a larger,
carefully arranged structure supports it.
6. Developing methods for treating a spherical cow is important; however, tailoring these for polygonal or real cows holds equal or greater significance.
7. If we need to categorize based on partial observation or estimation with uncertainty, we have to learn to accept the error.
8. PhD memes are the antidote to the impostor syndrome.
9. Cat propositions are neither opposable nor defendable, and thus not approvable.
10. In the journey of pursuing a PhD, one must have an addiction, among which sports is the healthiest option.
These propositions are regarded as opposable and defendable, and have been approved as such by the promotors Prof.dr. M. M. de Weerdt, and Dr. T. Abeel.
                           
      
  
            