Gym Workout IMU Dataset

Dataset Overview

This IMU (Inertial Measurement Unit) dataset has been recorded during various strength training exercises at the gym. It is designed to develop machine learning models that can accurately detect the type of exercise being performed and count the number of repetitions using data from an IMU sensor worn on the wrist.

Recording Details

1. Each file represents one set of a strength training exercise.
2. File Naming Format – Each file is named using the format:
   “ddmmyy_CODE_Wxx_Sx_Rxx”
   • ddmmyy: Date of recording
   • CODE: Unique exercise code (listed in a separate CSV file)
   • Wxx: Weight used
   • Sx: Set number
   • Rxx: Number of reps performed
3. Labeled Data – The dataset contains two label columns:
   • “Activity”: Exercise name abbreviation
   • “ActivityEncoded”: Encoded numeric representation of the activity
4. Sensor Specifications – Data was recorded using the Sensor Logger App on an Apple Watch SE at 100Hz frequency.
5. Sensor Placement – The IMU sensor was positioned on the left wrist for all exercises.
6. Total Sets Recorded – The dataset contains 164 sets of recorded exercise data.
7. Data Preprocessing Note – The first and last 1.5 seconds of each file may contain sensor lag and should be removed for accurate analysis.

Exercises Included

The dataset includes multiple strength training exercises, each labeled with an abbreviation:

• Upper Body Exercises
  • Straight Bar Lat Pulldown (SBLP)
  • Close Grip Cable Row (CGCR)
  • Neutral Grip Cable Row (NGCR)
  • Single Arm Pulldown (SAP)
  • Alternating Incline Dumbbell Bicep Curl (AIDBC)
  • Machine Preacher Bicep Curl (MPBC)
  • 30° Incline Dumbbell Bench Press (30DBP)
  • 75° Dumbbell Shoulder Press (DSP)
  • Dumbbell Lateral Raise (DLR)
  • Single Arm Cable Lateral Raise (SACLR)
  • Machine Rear Fly (MRF)
  • Face Pulls (FAPU)
  • Straight Bar Cable Tricep Pushdown (SBCTP)
  • Machine Shoulder Press (MSP)
  • Machine Tricep Extension (MTE)
  • Pushups (PUSH)
  • Pullups (PULL)
  • Assisted Pullups (APULL)
  • Preacher Curls (PREC)
  • Cable Rear Delt Pull (CRDP)
• Lower Body Exercises
  • Seated Hamstring Curl (SHC)
  • Smith Machine Squat (SMS)
  • Leg Extension (LE)
  • Standing Calf Raise (SECR)
  • Slow Half Smith Squats (SHSS)
  • Seated Calf Raise (STCR)
  • Isometric Leg Extension (ILE)
  • Hip Thrust (HT)
  • Lying Hamstring Curl (LHC)
  • Slow Single Leg Half Squat (SSLHS)
• Tricep & Bicep Workouts
  • Single Arm Overhead Cable Tricep Extension (SAOCTE)
  • Single Arm Overhead Dumbbell Tricep Extension (SAODTE)
  • Close Grip Overhead Cable Tricep Extension (CGOCTE)
• Chest & Shoulder Workouts
  • Machine Incline Bench Press (MIBP)
  • Machine Shoulder Press (MSP)
  • 45° Incline Dumbbell Bench Press (45DBP)
  • 30° Incline Bench Press (30BP)
• Wrist & Forearm Workouts
  • Dumbbell Wrist Curl (DWC)

Advantages of the Dataset

• Versatile Use Cases – Ideal for machine learning models, activity recognition, and fitness tracking applications.
• Highly Detailed Labels – Provides exercise type, set number, repetitions, and weight used.
• High-Frequency Data – Collected at 100Hz, ensuring precise movement analysis.
• Real-World Gym Data – Recorded in a natural gym environment with real workouts.
• Sensor-Based Insights – Helps in biomechanical analysis and performance tracking.

This dataset is perfect for fitness research, wearable technology development, and AI-based exercise recognition systems.