Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/ilirosmanaj/detect_kermit/llms.txt

Use this file to discover all available pages before exploring further.

kermit_model_evaluation.py is the main command-line entry point for the Detect Kermit project. It loads the trained ResNet model and runs predictions on either a single image, a comma-separated list of images, or a full video file. The script uses Python’s asyncio event loop internally, so predictions across multiple images or video frames are dispatched concurrently.

​
Synopsis

python kermit_model_evaluation.py [--file_type TYPE] [--files FILES]

​
Flags / Arguments

​
--file_type
string
default:"video"
The type of input to classify. Accepted values: image or video.
​
--files
string
default:"MuppetsEpisode3.avi"
Path to the file(s) to classify. When --file_type is image, accepts a comma-separated list of image paths. When --file_type is video, accepts a single .avi video path.

​
Short flag aliases

Short flagLong flag
-t--file_type
-f--files

​
Examples

python kermit_model_evaluation.py -t image -f kermit.jpeg

​
Output behaviour

​
Image input

When --file_type image is used, the script prints class probabilities to stdout for each image supplied. For example: 
Predicting the kermit.jpeg image
{'kermit': '99.87%', 'no-kermit': '0.13%'}
Each image in a comma-separated list is predicted independently and its result dict is printed in sequence.

​
Video input

When --file_type video is used, the script:
  1. Opens the .avi file with OpenCV and calculates the total duration in seconds.
  2. Extracts one frame per second (at 1 000 ms intervals) and saves each frame to episode3_results/ as a JPEG: 
    episode3_results/ep3_frame0.jpg
episode3_results/ep3_frame1.jpg
...
episode3_results/ep3_frameN.jpg
  3. Runs predict_image concurrently on every extracted frame.
  4. Writes a text banner directly onto each saved JPEG showing the prediction results, for example: 
    kermit 99.87% no-kermit 0.13%
    The overlay is rendered in red using cv2.FONT_HERSHEY_PLAIN at position (130, 25).
A progress percentage is printed to stdout while frames are being gathered from the video.
If the -f flag is omitted entirely, the script defaults to looking for MuppetsEpisode3.avi in the current working directory. Make sure you run the script from the repository root, or pass the full path explicitly with -f /path/to/your/video.avi.

Build docs developers (and LLMs) love

Get started for freeTalk to us