A PyTorch-based Library for Unsupervised Image Retrieval by Deep Convolutional Neural Networks

- Need someone with expertise in Deep Learning(CNNs) and ML to implement an Image Retrieval using NN mtech thesis work which is partially completed

- Base work already done, db - Caltech101, 9,900 images with 101 classes. VGG16, VGG19, Resnet, Inception & Exception already implemented in Google colab(python)

- Work on Novel technique(own method) required. Presently ResNet with XGBOOST has been implemented partially

- Literature survey has been done partially. (More common paper references and 3-4 lines on each paper required. Can be taken from any usual papers)

- One base paper finalised. Second base paper(ieee) also required commensurate to novel technique

- Vetting of existing codes, modifications if required, culling out results in form of graphs, charts, intermediate results, hyperparameter optimization techniques, comparision etc of all techniques reqd.

- Critical analysis of own(novel) technique vs base papers and projection of own method reqd

- Written work 80% completed, stuck at novel technique. All the analytics charts, graphs automatically retrieved from codes and some theory needs to be written in draft report.

- Anyone having worked on similar subject and is confident on Image retrieval or Computer Vision from academic point of view, can assist is implementing above codes and theory and also guide to defend the work required.

- Time period 7-14 days(main work already done, expert refinement and tweaking required)

Thanks and Regards