Skip to main content
34 votes
Accepted

What is the difference between semantic segmentation, object detection and instance segmentation?

Object Detection : is the technology that is related to computer vision and image processing. Its aim? detect objects in an image. Semantic Segmentation : is a technique that detects , for each pixel ,...
Blenz's user avatar
  • 2,084
8 votes
Accepted

Does image's background matter for detector training (CNN)?

Of course, it matters. Which one is best completely depends on your problem. The golden rule for machine learning problems is that you want the data you train on to be as representative as the data ...
Valentin Calomme's user avatar
7 votes

mAP scores on tensorboard (Tensorflow Object Detection API) are all 0 even though the loss value is low

I recommend several checks to make sure you get reasonable mAP@IoU scores for object detection API: Try varying the Intersection over Union (IoU) threshold, e.g 0.2-0.5 and see if you get an increase ...
Vadim Smolyakov's user avatar
5 votes

Unsupervised image segmentation

Fast answear Mean Shift LSH which is an upgrade in $O(n)$ of the famous Mean Shift algorithm in $O(n^2)$ well know for its image segmentation ability Some explanations If you desire a true ...
KyBe's user avatar
  • 420
5 votes
Accepted

Creating a Object Detection model from scratch using Keras

Before I answer your question, let me tell you this, You can go on and train a model from scratch, but you will definitely end up using one of the object detection architectures, be it Mask R-CNN, ...
praisethemoon's user avatar
4 votes

What is the difference between tensorflow saved_model.pb and frozen_inference_graph.pb?

saved_model.pb may represent multiple graph definitions as MetaGraphDef protocol buffers. Weights and other variables usually aren't stored inside the file during training. Instead, they're held in ...
Rohit Sharma's user avatar
4 votes

Best neural network architecture for object detection without location

It mostly depends on the amount of data you have available for training. However, unless you have an extremely limited dataset and you are unable to generate synthetic data, CNNs are currently the ...
Ben's user avatar
  • 2,572
4 votes
Accepted

Does resizing images during training affect the bounding box annotations?

My doubt is whether the resize would now change the position of my object according to annotation? Yes, it will. should i annotate on the resized images(resize them myslef before training?) No, ...
Simon Larsson's user avatar
4 votes

what is darknet and why is it needed for YOLO object detection?

https://pjreddie.com/darknet/ is their website... I cite : "Darknet: Open Source Neural Networks in C Darknet is an open source neural network framework written in C and CUDA. It is fast, easy to ...
Lucas Morin's user avatar
  • 2,289
3 votes
Accepted

What is difference between intersection over union (IoU) and intersection over bounding box (IoBB)?

The Intersection over Bounding Box is the Intersection over Union (IoU) for object detection tasks, where you have a bounding box. There are many tasks (e.g. image segmentation) where you have an IoU ...
JkBk's user avatar
  • 432
3 votes

Unsupervised image segmentation

You may need to take a look at this work submitted and accepted for CVPR 2018 : Learning to Segment Every Thing In this work, they try to segment everything, even objects not known to the network. ...
LeNoir's user avatar
  • 31
3 votes

How can I read input node and output node from .h5 file in keras model?

I assume that you have the Keras .h5 file ready. Now, we need to load the model from that file using, model = keras.models.load_model( 'model.h5' ) ...
Shubham Panchal's user avatar
3 votes
Accepted

How can we extract fields from images?

I have a similar use-case and a working product based on tensorflow object-detection api and pytesseract for OCR. On top of the extracted text, I perform regex for validation of the extracted ...
loki's user avatar
  • 184
3 votes
Accepted

Papers presenting results that are worse than random chance

Not all probabilities are 50/50. I'm assuming that the paper you are looking at is eYOLOv3-Lite: A Lightweight Crack Detection Network. Under the evaluation metrics section, I see For each image,...
Brady Gilg's user avatar
3 votes

Which is the "BEST" deep learning model for "Custom" object detection for images & real time. YOLO v3, v4, v5, EfficientDet?

There are many subtle differences between yolov4 and v5 other than speed like YOLOv4 exceeds YOLOv5's performance on the COCO benchmark.this link will help you explain. YOLOv4 VS YOLOv5. I dont know ...
Shiv's user avatar
  • 689
3 votes
Accepted

Loss decreases, but Validation Loss just fluctuates

It looks like your model is overfitting: it's learning from the training dataset, but this learning doesn't apply to the test dataset. You can try to reduce the complexity of the model by simplifying ...
alexmolas's user avatar
  • 566
3 votes

What is difference and factors that help to make a decision between choosing training using pre-trained model vs from scratch?

Does pre-trained model in the above quote mean that it will re-use the training data from the pre-trained model (I can't see the original files used in the pre-trained model), plus my 300 images (100 ...
noe's user avatar
  • 27k
3 votes
Accepted

what is the difference between 'object detection' and 'outlier detection' in computer vision?

Object Detection: Labelling an sample to indicate the presence of a class of object. In images, this is usually done with a bounding box or applying labels to each pixel of the image. Outlier ...
James Ashford's user avatar
3 votes
Accepted

Spatial Join Pandas Dataframes of Bounding Boxes (cross match)

Edited after comments in other answers: Generally speaking, I would reframe this as a linear sum assignment problem. This can be solved using a modified version of Munkres algorithm allowing a cost ...
Just trying's user avatar
3 votes

Car Make and Model detection

For detecting the make and model of cars from images with high accuracy across a large number of classes, I would recommend a convolutional neural network (CNN) architecture tailored for fine-grained ...
Multivac's user avatar
  • 2,999
2 votes

Unsupervised image segmentation

The state-of-the-art (SOTA) for image segmentation would be Facebook's Mask-RCNN. While it is usually trained on dataset such like COCO or Pascal which feature real-life objects, you can re-trained ...
Arthur Douillard's user avatar
2 votes

Unsupervised image segmentation

Actually, your task is supervised. Segnet can be good architecture for your purpose which one of its implementations can be accessed here. SegNet learns to predict ...
Green Falcon's user avatar
  • 14.1k
2 votes

Best neural network architecture for object detection without location

May not really answer you question, but to be honest, if you just want to find one specific object that you know beforehand, I wouldn't use a neural network at all (in my opinion you should use the ...
nyro_0's user avatar
  • 153
2 votes

Object Detection classification

In general, you want to show your model all types of "resistors" it would see in the real world. Is it feasible for you to manually go through your dataset and label each of these images as resistor ...
Andy M's user avatar
  • 400
2 votes
Accepted

Space between an object and the ground truth bounding box

The first image is a better bounding box. A perfect bounding box has no space in between the edges of the object and the box, but it also fully contains the object.
Ben's user avatar
  • 2,572
2 votes

How to label "other" while labeling image for object detection/classification?

This highly depends on your test data. Suppose you have trained your data which all contains some kind of food. If you give it a hand at test time, it will try to find the most similarity it has with ...
Green Falcon's user avatar
  • 14.1k
2 votes
Accepted

Very slow convergence with CNN

Implement the below mentioned techniques and check Add Batch Normalization Increase the learning rate Standard/Normalize the inputs if you have not done it already
Bharath Kumar L's user avatar
2 votes
Accepted

Issues with training SSD on own dataset

What you are experiencing is called overfitting and it happens because of your very small dataset. All the model cares about is performance on the training dataset, so given the opportunity, it will ...
Mark.F's user avatar
  • 2,230
2 votes
Accepted

Extracting the feature map in Tensorflow Object Detection API

...
yildirim's user avatar
  • 168
2 votes
Accepted

Preparing custom dataset for object detection using ML

I have the following suggestion: The size of your data-set is small, you should increase the sample size. You also have to use data augmentation methods for improving the performance, translation, ...
Green Falcon's user avatar
  • 14.1k

Only top scored, non community-wiki answers of a minimum length are eligible