Manu Chopra

Summer of 2014, I travelled to Qingling Village in Chongqing Province, China and constructed two bridges and renovated the sanitation system for the local school along with five other friends from Stanford School Of Engineering.

As a part of the class EE17N at Stanford, We (a team of 3 people including me) designed flexible chips by testing the quality of several non-conductive flexible substrates using a polyamide base and copper and gold-chromium metal circuits. The project introduced us to the brilliant Stanford Nano Fabrication lab. We succeeded in creating a low cost solution to create such flexible patterns for wearable devices.

When we are out with our friends late at night in Indian metropolitans, are we safe? Let’s face it: we have made peace with the level of crime (specially against women) in our society and don’t blink an eye when we hear of a new incident. We live with eve-teasing and hence even embolden attackers.

Awaaz is a non-profit whose goal is to put an end to this hypocrisy: a culture that worships goddesses disrespects its women. The team's first goal is to empower women to avoid any such… Show more

When we are out with our friends late at night in Indian metropolitans, are we safe? Let’s face it: we have made peace with the level of crime (specially against women) in our society and don’t blink an eye when we hear of a new incident. We live with eve-teasing and hence even embolden attackers.

Awaaz is a non-profit whose goal is to put an end to this hypocrisy: a culture that worships goddesses disrespects its women. The team's first goal is to empower women to avoid any such scenarios where we’d be at risk. Using a layer of technology on top of proliferating smart devices, and the youth’s desire to be safe and free, this project aims at developing crowdsourced safe routes for our mundane commutes. Show less

Imagined speech refers to the process in which a subject imagines speaking a given word without moving any muscle. We classified the syllables “ba”, “ku”,“im” and “si” from imagined speech EEG signals. Our neural networks model is successfully able to classify syllable pairs from the EEG data with over 90 percent accuracy.

With the amazing advancements in the recent tide in image recognition, we are able to piece together technologies in multiple knowledge areas to begin to tackle the complexity of gathering information in videos in a broad and useful manner - namely with text. This was made firstly possible by the research success of Stanford researchers Andrej Karpathy and Fei-Fei Li who managed to train images to describe the scenes depicted in them in a technology dubbed NeuralTalk. Image recognition has made… Show more

With the amazing advancements in the recent tide in image recognition, we are able to piece together technologies in multiple knowledge areas to begin to tackle the complexity of gathering information in videos in a broad and useful manner - namely with text. This was made firstly possible by the research success of Stanford researchers Andrej Karpathy and Fei-Fei Li who managed to train images to describe the scenes depicted in them in a technology dubbed NeuralTalk. Image recognition has made steady leaps in classification accuracy, but there finally seemed a way to capture verbs and deduce actions.

Our hack uses this technology to describe successive frames of a video and then match the best one with a query. After sampling the video, the images are first represented using a 16-layer convolutional neural network which attains a high-level “strong” representation. They are then transformed into a “modular space” that allows the representations to be unfolded in a long short-term memory recurrent neural network that maps to a language model. Lucky for us, great pretrained models exist for all these architectures!

Incorporating the successful object recognition capacities of the Imagaa API, we now only had to match our extracted information to a user’s query. Here we used the MetaMind semantic similarity feature which compared the query with the annotated sentence. By analyzing the contexts of the sentences, their word vectors, synonyms, syntactic roles - basically everything short of dark magic - MetaMind’s semantic similarity allows us to match the neutral delivery capacities of NeuralTalk with a diverse set of user queries. If any of the frames met a threshold for acceptable level of matching to the query, we accepted it. Otherwise we report the query not found.

TL;DR. A preprocessing step and many more neural networks later - like magic - scenes in a video are represented with text and compared to a search.
Show less

Created a score-based accountability system that incentivizes health workers in Kano, Nigeria to complete daily tasks, refresh their medical knowledge, and achieve a high standard of engagement in their work. The Android based App was built in partnership with eHealth Africa, a nonprofit organization based in Nigeria and EE 46 - Engineering For Good to help transform the Nigerian healthcare sector.

Imagine a box that has everything you need. A box that connects your physical world to the virtual world. A box no more or less ordinary than the text box I am writing this description on. A box that understands what 'you'​ want and gives it to you within seconds. A box that connects you to everything you need. We are introducing the Smart Box, it's everything you need and even more in one simple text box.

Wanna play that Beyonce song? Just ask the box. No need to open Spotify again… Show more

Imagine a box that has everything you need. A box that connects your physical world to the virtual world. A box no more or less ordinary than the text box I am writing this description on. A box that understands what 'you'​ want and gives it to you within seconds. A box that connects you to everything you need. We are introducing the Smart Box, it's everything you need and even more in one simple text box.

Wanna play that Beyonce song? Just ask the box. No need to open Spotify again. Hungry for pizza? You can ask the box to bring you food. Looking for stuff to buy/ sell on eBay? Ask the box. The box is fueled by some amazing services including Google's Freebase API. You can search for your favorite pictures from Flickr and, just like a friend, you can talk to the SmartBox. But the real magic begins with the box's entry to the physical realm.

Who is our target user? Everyone on the internet and everyone who will be on the internet in the future. SmartBox was designed to connect every internet user to developers who design delightful experiences. Our team was able to achieve some amazing features which we are incredibly proud of.

API Architecture allows both voice and text commands in a modular way which allows for future expansion and easy access. Our software architecture allows us to seamlessly integrate several services across the Internet.
Voice recognition software. Annyang allows us to continuously process both voice and text and use the processed media to create an amazing user interface.
An amazing collection of APIs - We are honored to be a part of such an amazing community where developers create awesome software for everyone to use. These APIs allow us to make such hacks.
A hardware that measures your brain activity through a 3D-printed bracelet.
All of this in a seamless, simple and friendly software: everything you need is just one box away from you, literally.
Show less

TreeID is an automatic leaf image recognition system for plant species identification. The aim of the project is to classify all of the 60,000 trees on Stanford’s campus. The system is based on a Growing Convolutional Neural Network model, in which the network grows up itself until it solves the target problem thereby, achieving the best trade-off between classification accuracy and computation cost.