EVA-4

This repo contains all the assignments from the course called EVA conducted by the 'The School Of AI'. Object detection, Object reconisation, segmentation, monocular depth estimation.

1. Basics of Python - link
2. Setting Up Basic Skeleton - link
3. Pytorch Basics - link
4. To achive 99.4% validation accuracy for MNIST DatSet - link
5. The whole process- Coding drill - link
6. Regularisation techinques on MNIST - link
7. Advanced convolutions(depthwise seperable and dialated convolutions) on CIFAR 10 - link
8. 85%+ Validation accuracy target for CIFAR 10 -link
9. 87%+ Validation accuracy target using Albumentation library for agumentation. -link
10. 88%+ Validation accuracy target using Reduce Lr On Pleateau and Implentaion Of Lr finder - link
11. Introduction Of SuperConvergence and 90%+ accuracy for CIFAR dataset using One Cycle Policy - link
12. Prediction best no of bounding box for 50 dog images and Training TinyImageNetData on RESNet 18 and achive 50%+ accuracy. link
13. YoloV3 implementation on Custom created dataset via transfer learning - link
14. Dataset Creation for Monocular depth Estimation and Segmentation - link
15. Final Assignment - Monocular depth Estimation and Segmentation - link

PytNet - Contains all the functions used for these assignment. Custom library built on top of Pytorch. - link