DevoNet I have started working on the devonet model as per my gsoc proposal ,As we are using cell tracking dataset and dataset we have the 3d images of with nucleus detection and nucleus segmentation process whole data is in the format of TIF format and images are grey scale , labels and detection.
Here in this week completely we worked on image encoder part , where orginal images as sam model is trained on the images of color data , the data need to shifted to the original decoder accepted shape .
Here in this week most of time went on making the data for the prompt encoder as it accepts the prompt embeding , I have to convert all masks into bounding boxes.
Gsoc coding period week 1 have started coding and setup is complete the sam model is heavy and can,t be trained on local computer ,so most of my work will be using kaggle gpu.
Finally ,after 5 months of consistent work and sleepless night the d-day has approached ,continous doubts started poping my head and was really confused that , have given my best shot.