Key takeaways
- The Computer Vision Classic hackathon attracted over 200 registrations and nearly 90 active participants.
- V G Sravan, a third-year student, utilised his passion for Deep Learning to excel in the competition.
- Sravan achieved top scores by employing ResNet50 and optimising his model with specific tweaks for better accuracy.
- The hackathon provided a welcoming environment for beginners interested in Machine Learning and data science.
MachineHack successfully concluded its eleventh instalment of the weekend hackathon series last Monday. The Computer Vision Classic hackathon provided the contestants with an opportunity to revisit the very basics of Computer Vision by solving the classic Cats & Dogs problem.The hackathon was greatly welcomed by data science enthusiasts with over 200 registrations and active participation from close to 90 practitioners.
Out of the 87 competitors, three topped our leaderboard. In this article, we will introduce you to the winners and describe the approach they took to solve the problem.
#1| V G Sravan
Sravan is a third year Electronics and Communication Engineering student at IIT Kharagpur. Intrigued by the technology and its advancements, Sravan started his journey towards Data Science and Machine Learning with his focus set on Deep Learning.
Amidst the pandemic, Sravan had been using his time in practising Machine Learning by solving online hackathons.
“MachineHack is a very good place for beginners, it’s a very easy and competitive place to work on different ML models” - Sravan shared his opinion about MachineHack
Approach To Solving The Problem
Sravan explains his approach briefly as follows:
This hackathon is a rather very basic yet model-driven hackathon. I used ResNet50 with ImageNet pre-trained weights as my main architecture, after trying out various other architectures like ResNet34, 101, VGG_16 .etc
Even a very small and simple model could produce above 95 percent accuracy but the tweaks which I made took me to the top of this competition’s leaderboard (tweaks are explained in the code itself). I also took advantage of the vague dataset which contained images of different resolutions and aspect ratios
Get the complete code here.
#2| P V Sai Krithik
Sai Krithik is a B-tech student currently in his second year of Electronics And Communication Engineering at Vidya Jyothi Institute of Technology ,Hyderabad.
He had his first encounter with machine learning during his first year of college in an NLP hackathon conducted at IIIT Hyderabad. The hackathon provided his team a great learning experience and gave him a passion to continue his journey towards the demanding field. He started with the basics of ML and dedicated his time for doing hands-on projects.
“I have been participating in the weekend hackathons since the beginning, I get good public scores every week and it is always very competitive there.I get to connect with experts and gain more knowledge.”- he shared his opinion.
Approach To Solving The Problem
He explains his approach briefly as follows:
In this hackathon, I started with the Convolution Neural Networks using keras-tensorflow which didn’t give a good accuracy score. Then I went on to try different transfer learning models like ResNet152, DenseNet201, ResNet101 and ResNet50. ResNet50 got me a good validation score, I trained the model for 4 epochs with an observed learning rate and again trained 1 more epoch with a lower learning rate and I was able to achieve 0.99788% accuracy in my validation which gave a private score of 0.99803% accuracy.
Get the complete code here.
