Project number: #14
Title: Android based Object Detection and Classification: Modeling a Child's Learning of What's Hot and Cold

*************************************************************

Should this project be considered for the Best Project award? no

Should this project be considered for the top 3 project awards? no

On a scale of 1 to 10, how would you rate the overall organization/clarity of the project report? 8

5. On a scale of 1 to 10, how would you rate the overall project idea? 7

6. On a scale of 1 to 10, how would you rate the overall research contribution of the project idea, methodology and/or results? 8

* Overall, is the project report clear, concise, and well-organized?
Yes it was well organised.The explanation of the usecases and how the code is implemented is clean. 

* How does the project idea and methodology fit within the framework of Developmental Robotics?

The original proposal or the idea of trying to learn about the hot/cold or other charecterestics of an object is something interesting and would be helpfull for the field of developmental robotics.At the same time its a difficult task.Hence from the results and the report the author accepts that he was not able to actually impart the learning of hot or cold objects.Whatever is done is an implementation of image processing(which is essential as the images serves as visual inputs) and learning using SOM.

* Describe what you like BEST about the project?

The idea is something good and the author has tried using SOMS and android platform to deploy the project which should have involved a considerable amount of work.

* Describe what you like LEAST about the project?

The idea is very ambitious. Giving the paper a title which is something not done inside the paper in the implementation.The datasets or tests are done with a pen,eraser and other objects which is no way realted to being hot or cold.But whatever is done by ther author is only doable in the available time and really understanding whats cold and hot would take lot more time and research.

* Do the methods, results and contributions of the final project
correspond to what was presented in the initial project proposal?
Not totally. The proposal has a wider scope as mentioned.

* Are there any major details left out with regards to the methods,algorithms, or experimental design described in the report?
The algorithms and the methodology used to implement and the conversion between various libraries in various devices and the constraints of each have been well explained in the paper.The only problem is that the project does not totally implement what was originally proposed.

* Do the experimental results reported in the paper demonstrate success?
Not totally.The system is successfull in image identification and predicting using SOM.But this is not what was supposed to be success.If the author has actually demonstrated some test using an object as icecream or pizza which was originally mentioned in the paper and showing some realted objects and trying if there is any relationship spotted based on sensors readings.


* Do you have any suggestions for improvement and future work?
Future works are well laid out at the end of the paper by the author himself. I am not quite convinced about just using sensors and image details for exploring the properties of an object. This is because if the idea is just to know about the temperature then sensors to measure the temperature are just fine.As mentioned in the initial review the object itself has the property of hot or being cold and this task of seperating a container from the content is something difficult.Humans are introduced to the containers along with the content but we get a chance to learn about just the container and the content seprately. THis means there is a seperate learning involved for both the containers and understanding its properties and understanding its content which has some properties. Also one thing to note here is that the objects properties vary over time. Say the icecream is cold only for sometime or a hot object lose the heat after a while. So we don not really corellate the hotness or coldness blindly to an object which is obtained by visual inputs rather we do it by using other sensory inputs like touch at times. But we also capable of observing the vapors out of an object to identify if it is hot or not even without touching it.This is something which might not be feasible by image processing. Also the opencv library provides libraries for image processing.This would not be able to identify an object if the ligh setttings or RGB values differ slightly which becomes a great constraint when it comes to food or related stuff as the colors could be totally varying each time. So the algorithm might consider something which do not consider the color of the object as a dominant property for object identification.

There was a mention that "Each set would be independently tested to see the impacts of color on determining if a object is going to be cold/hot". I am not sure how this would work as color does not mean anything to humans when it comes to hot or cold. We could realize such temperature differences with no realtion to color. Also in real time objects in any color could have such properties. So considering color as an important property seems to be a not very good approach to me.

Also I was just wondering what an object would be categorised if it is neither hot nor cold.Because such objects occur in reality.I am not sure what the pen would be considered as by the system if it should categorize as hot and cold.


* How close is the final project report to being publishable as a conference or journal paper (consider the research papers that were
part of the course reading)? What would it take to get there?

The project might not be to the standards of getting published as there is not much novel in terms of what is implemented.It is something new in terms of implementation,say using Android for doing learning or running algorithms for image matching/recognition.This is a project which is advanced in terms of technology used or implemntation but not in terms of concept or idea.To make this publishable the project would have to do what the title suggests.