hillarymutaik

Hillary Mutai's repositories

End-To-End 6D Object Pose Estimation in Crowded Scenes The solution to the OPE using the CLUBs dataset and the python proramming language for the artifficial intelligence and machine learning. Python have strong libraries for these disciplines mentioned. To name a few, the Tensorflow is used for extensive research into machine learning and artifficial intelligence. It is one of the strong libraries in python for these purpose. In this project, Tensorflow usage will be determining the different kind of images and their appearances. This is one of the strong grip in the newural network for classification of objects in 6D. Another library useful in the pandas. This is a very useful in reading of datasets, making data wrangling and regression and analysis. Numpy does the supportive to this training and testing of data. Matplotlib and plotly are major python librarioes for plotting of the analyzed data, for presentation and either simulation. Last but least is the use of the OpenCV, a powerful python library used for image detection and recognition. In this project, these python libraries will be highly used in the object pose estimation. The CLUBS being the dataset used, for classification, segmentation and detection, any crowded objects will be analyzed usin the above libraries to accomplish the targeted goals and the two possible solutions: End-to-end direct pose regression usin deep learning, that is Efficient Pose, DPOD and,Network with multiple parts that first extracts features. The Scaling of the 2D ojects in bounding boxes and pixel-wise labels in the RGB images. With the data containing approximately 85 objects and 25 box scenes, different box scenes have different configurations of objects in them vary. The RealSense cameras , callibrations of two different files provided , one containing higher-resolution depth and the other with lower resolution depth. The best approach in working in the project is using an hybrid of the already existing approaches, technics and that can lead to developiong a totally a new approach too. This is analysing the existing methodologies and finding their weaknesses and find the best solution from it.

Language:Python300

admin-panel

admin panel

Language:DartMIT200

Event-Tracker-App

Inventory App - Android Mobile Application

Language:JavaMIT200

airduka

Single page flutter app that would be deployable on both Android and IOS. The app is expected to make a request to the provided API then return a list of products where one can be able to switch/toggle between List view and Grid View without recalling the API in that process.

Language:Dart100