Code for the paper "Modeling Multiple Temporal Scales of Full-body Movements for Emotion Classification"

No description available

A machine learning project that classifies news articles as real or fake based on their content (title, body text, author, source, etc.). This project demonstrates the full pipeline of a binary text classification task using various NLP techniques and ML algorithms.

Real-time face, pose, and hand detection with OpenCV, MediaPipe, and DeepFace for age, gender, emotion, race, and celebrity look-alike analysis

This is a AI security camera made with Raspberry pi. It will send you the Email alert whenever it detect the Human being. You will also have the option of life streaming through internet. You will have three option of classification ie. Full body detection, Upper body detection & Facial detection

Multiclass Multilabel Prediction For StackOverflow Questions Data set : https://www.kaggle.com/therajeshreddy/stackoverflow Objective : Given text for Questions from StackoverFlow posts, predict tags associated with them. This is a scaled down version of predecting only top 10 most occurring tags Programming Language : Python using nltk & Keras Model Architecture : Deep Learning using Recurrent Neural Network (RNN) About Data Set Dataset has text of questions, answers and thier corresponding tags from the Stack Overflow programming Q&A website. This is organized as three files: Questions contains the title, body, creation date, closed date (if applicable), score, and owner ID for all non-deleted Stack Overflow questions. Tags contains the tags on each of these questions. Answers contains the body, creation date, score, and owner ID for each of the answers to these questions. The ParentId column links back to the Questions table. We don't use this file as we want to predict Tags given a question Data Pre-Processing Questions File Code : Stackoverflow Clean Questions.ipynb Read Questions File Drop All columns except Id,Title and Body Now the text in the Body column seem to have many html tags in the text. We use Regular Expressions and Clean the Body column text by removing the html tags import re def rem_html_tags(body): regex = re.compile('<.*?>') return re.sub(regex, '', body) ques['Body'] = ques['Body'].apply(rem_html_tags) Save the questions file for later use ques.to_csv('question_clean.csv',index=False) Tags File Code : Stackoverflow Tags Map & Model.ipynb Read Tags File Identify top 10 Tags by count tagCount = collections.Counter(list(df_tags['Tag'])).most_common(10) print(tagCount) [('javascript', 124155), ('java', 115212), ('c#', 101186), ('php', 98808), ('android', 90659), ('jquery', 78542), ('python', 64601), ('html', 58976), ('c++', 47591), ('ios', 47009)] Manipulate the tags dataframe so that all the Tags for an ID are as a list in a row (grouped by Question ID) def add_tags(question_id): return tag_top10[tag_top10['Id'] == question_id['Id']].Tag.values top10 = tag_top10.apply(add_tags, axis=1) Combine the Questions and Tags Code : Stackoverflow Tags Map & Model.ipynb Merge the Questions and Tags data frame by ID total=pd.merge(ques, top10_tags, on='Id') Our Dataset would now have only Id, Title, Body & Tags Text Preprocessing Code : Stackoverflow Tags Map & Model.ipynb We will use nltk, preprocessing from Keras and sklearn to process the text data Tags preprocesing Use MultiLabelBinarizer from sklearn on the Class labels(Tags) from sklearn.preprocessing import MultiLabelBinarizer multilabel_binarizer = MultiLabelBinarizer() multilabel_binarizer.fit(total.Tags) print(multilabel_binarizer.classes_) array(['android', 'c#', 'c++', 'html', 'ios', 'java', 'javascript','jquery', 'php', 'python'], dtype=object) Title & Body Preprocessing Tokenize the words Convert the tokenized words to sequences Model Building Implemented a Hybrid model in TensorFlow using Keras as high level api. Architecture used is RNN. In this model first we train a model using the Title data, then train a model using the Body data. Outputs of both are concatenated and passed thorugh the dense layers before connecting to the output layer RNN Model : The model first uses GRU for the sequence data training with 2 GRU layers one for Title and other for Body. RNN for Title has 1 Embedding Layer has input of Title vocabulary length(68969) + 1(for 0 padding) and out put of 2000 embeddings (for better results use full vocabulary length+1) 1 Gated recurrent unit (GRU) layer 1 dense output layer of shape 10(No of classes(tags) we are trying to predict) # Title Only title_input = Input(name='title_input',shape=[max_len_t]) title_Embed = Embedding(vocab_len_t+1,2000,input_length=max_len_t,mask_zero=True,name='title_Embed')(title_input) gru_out_t = GRU(300)(title_Embed) # auxiliary output to tune GRU weights smoothly auxiliary_output = Dense(10, activation='sigmoid', name='aux_output')(gru_out_t) RNN for Body has 1 Embedding Layer has input of Title vocabulary length(1292018) + 1(for 0 padding) and out put of 170 embeddings (for better results use full vocabulary length+1) 1 Gated recurrent unit (GRU) layer # Body Only body_input = Input(name='body_input',shape=[max_len_b]) body_Embed = Embedding(vocab_len_b+1,170,input_length=max_len_b,mask_zero=True,name='body_Embed')(body_input) gru_out_b = GRU(200)(body_Embed) Combine the 2 GRU outputs com = concatenate([gru_out_t, gru_out_b]) The fully connected network has 2 Dense Layers 1 Dropout layer 1 BatchNormalization layer 1 Dense Output layer # now the combined data is being fed to dense layers dense1 = Dense(400,activation='relu')(com) dp1 = Dropout(0.5)(dense1) bn = BatchNormalization()(dp1) dense2 = Dense(150,activation='relu')(bn) main_output = Dense(10, activation='sigmoid', name='main_output')(dense2) Model Compilattion with optimizer='adam', loss='categorical_crossentropy', metrics='accuracy') Model Performance Review Classification Report to check Precision, Recall and F1 Score The Model seem to performing good enough with score of 84%. Increase in the Embedding, GRU and dense layers would help in getting better results Random Validation on Test Data Save the Model & Weights Saving the model for transfer learning or model execution later model.save('./stackoverflow_tags.h5')

Full Body classification

Full-body IMU-based terrain classification using terrain-aware step segmentation and multi-branch CNN

This is a real-time system that uses advanced body pose detection and machine learning to identify A–Z alphabets from full-body gestures. Built with Flask, OpenCV, and MediaPipe, it offers classification and data collection modes, enabling accessibility, learning, communication, and gesture-based interactions

Implemented full-body image segmentation using the MADS dataset, focusing on accurate pixel-wise classification of human body parts. Leveraged deep learning models to segment regions such as head, torso, limbs, and clothing, enabling detailed human parsing for tasks like pose estimation and fashion analysis.

An NLP text-classification benchmark for detecting fake news from full-length articles (title + body). This content-only approach avoids social/network data to create a fair, deployable baseline that isolates textual veracity signals from contextual bias, outputting a binary fake (0) or real (1) label.

Nowadays, acoustic guitars are more prominent than any time in recent memory. With such a large number of incredible artist musicians and acoustic guitar players, it's no big surprise why. On the off chance that you are hoping to get into the diversion then you have to recollect that your guitar should feel like an augmentation of yourself, however finding that immaculate fit is no simple undertaking. With such a large number of various producers, styles, bodies and highlights out there, it can influence your make a beeline for turn. Remain back for a moment and take a full breath in light of the fact that before you buy your next acoustic guitar, make sure to take after these tips to ensure you discover the guitar that best suits you. Related: 10 Best Acoustic Guitars Under $500 + 10 Best Electric Guitars Under $500 1. Financial plan For the majority of us, this is the first (and maybe most imperative) thought. Like anything, your spending will decide a great deal about the nature of the instrument that you buy, yet there is a sure point where the profits truly won't make any difference unless you a genuine expert. Furthermore, all things considered, you most likely don't should read this guide! For novices, $300 is generally a really decent number. There are a lot of shabby acoustic guitars out there for between $75-$150 and keeping in mind that these can be enticing, as they may regularly incorporate units, for example, a conveying case, a learning book, and a stand, these sorts of bundle bargains are generally excessively great to be valid. These low-estimated guitars are typically very hard to play and sound terrible, so spend somewhat more and get something you'll appreciate playing. Rather, look at our rundown of the 10 best acoustic guitars under $500 for a couple of awesome decisions. For middle and propelled guitarists, about $700 is regularly a decent value point. Around here, you will locate the quality, solid acoustic guitars you are searching for. Now, the guitar ought to incorporate a spruce (wood) top and the plastics in there ought to be specifically identified with any hardware that you might search for. For the gatherer or devotee, surely there are a lot of astonishing and delightful acoustic guitars over $1,000. In case you're looking that range, you most likely needn't bother with this article. 🙂 2. Style There are a few diverse acoustic guitar styles out there. In the event that you are a learner, you may need a standard acoustic guitar, however there are a lot of other incredible choices out there. Acoustic guitars can be separated into three unique classifications: standard, traditional and fingerstyle. Every one of these sounds distinctive so give them a shot and see what fits you. There are focal points to every one of them. For instance, with a fingerstyle guitar , you have some magnificent playability for blues and jazz styles while you can get the shining Spanish sound with a traditional hatchet. 3. Acoustic-Electric Discussing gadgets, this is a choice you should as of now have made in the positive. There are some incredible straight acoustic guitars, however the handiness is more restricted. They expect you to have greater hardware on the off chance that you need to record (i.e. you will require a different receiver) and additionally to gigging in boisterous conditions. When choosing your acoustic-electric guitar, look at how it sounds snared to an amp. Despite the fact that you can play on about any enhancer, you will need to likewise buy a decent amp that is intended for acoustic guitars. These for the most part have brighter, clearer sounds than those intended for an electric guitar and will be justified regardless of the venture once you choose to begin playing live. They additionally can enable you to adjust your sound and utilize distinctive impacts, for example, reverb and reverberate so you can truly change your style. 4. Set-Up Acoustic guitars can change generally on how they are composed. For instance, there are battleship style guitars which are gigantic (thus the name) and have an uproarious, blasting sound. Then again, you acoustic-electric models that resemble burrowed out Fender Strats. Once more, this is controlled by what sort of player you are. Numerous acoustic guitars are composed with the goal that the neck joins that body at the twelfth fret making playing the higher notes more troublesome. In the event that you are hoping to shred as far as possible all over the neck, at that point you might need to search for a cut-out. On the off chance that you are more into a cadenced style than you might need to run with a more customary shape. In any case, make sure to discover a guitar that is agreeable for you to play so make certain to consider extra factors, for example, the activity and the fret buzz. 5. The Guitar For You A considerable lot of these tips could apply to a guitar you buy, acoustic, electric or even instruments like basses and ukuleles. There are some extra factors you will need to consider, for example, the materials it is produced using (don't kick us off on the diverse woods out there) and in addition distinctive "excellence marks" like fretboard decorates and outlines. Keep in mind, the most imperative things are the means by which it sounds and how it feels. Everything else is a reward! Diego has an enthusiasm for music since he was 12 years of age. Getting a charge out of sticking and instructing, he runs The Musician Lab and get included with music.